https://blog.schoergendorfer.com/post/orchestration-is-the-new-challenge-for-cx-in-the-age-of-ai-agentsSecFeedde2026-08-26T14:47:53.386980+00:00Orchestration is the new challenge for CX in the age of AI agentshttps://blog.schoergendorfer.com/post/cisa-vulnerability-reviewSecFeedde2026-08-26T14:32:43.160055+00:00CISA Vulnerability Reviewhttps://blog.schoergendorfer.com/post/what-would-have-to-be-true-for-agentic-coding-to-replace-junior-engineersSecFeedde2026-08-26T14:32:43.126540+00:00What Would Have to Be True for Agentic Coding to Replace Junior Engineershttps://blog.schoergendorfer.com/post/finops-for-the-ai-era-new-flexible-billing-and-cost-controls-for-agentsSecFeedde2026-08-26T13:47:40.166181+00:00FinOps for the AI era: New flexible billing and cost controls for agentshttps://blog.schoergendorfer.com/post/infrastructure-for-the-ai-era-dynamic-capacity-management-for-agentsSecFeedde2026-08-26T13:47:40.133363+00:00Infrastructure for the AI era: Dynamic capacity management for agentshttps://blog.schoergendorfer.com/post/ai-driven-software-delivery-with-kiro-aws-devops-agent-and-bluebox-by-dynatraceSecFeedde2026-08-26T13:18:15.151674+00:00AI-driven software delivery with Kiro, AWS DevOps Agent and Bluebox by Dynatracehttps://blog.schoergendorfer.com/post/edge-infrastructure-under-siege-what-two-independent-datasets-reveal-about-whosSecFeedde2026-08-26T13:18:15.113407+00:00Edge infrastructure under siege: what two independent datasets reveal about who's exploiting your perimeterhttps://blog.schoergendorfer.com/post/when-ai-agent-traces-become-application-dataSecFeedde2026-08-26T13:02:34.863824+00:00When AI agent traces become application datahttps://blog.schoergendorfer.com/post/usn-8678-2-openssl-openssl-10-vulnerabilitiesSecFeedde2026-08-26T13:02:34.830431+00:00USN-8678-2: OpenSSL, OpenSSL 1.0 vulnerabilitieshttps://blog.schoergendorfer.com/post/microsoft-just-released-agent-lightning-v10-heres-why-it-matters-for-platform-enSecFeedde2026-08-26T12:32:36.322281+00:00Microsoft just released Agent Lightning v1.0. Here’s why it matters for platform engineers.https://blog.schoergendorfer.com/post/making-room-for-whats-next-in-the-gitlab-uiSecFeedde2026-08-26T12:17:30.623939+00:00Making room for what's next in the GitLab UIhttps://blog.schoergendorfer.com/post/governance-guidance-for-cncf-projects-choosing-the-right-structure-for-your-proj-2SecFeedde2026-08-26T12:17:30.539414+00:00Governance guidance for CNCF projects: Choosing the right structure for your project’s size and stagehttps://blog.schoergendorfer.com/post/usn-8681-1-openjdk-25-vulnerabilitiesSecFeedde2026-08-26T12:02:51.750060+00:00USN-8681-1: OpenJDK 25 vulnerabilitieshttps://blog.schoergendorfer.com/post/exploits-and-vulnerabilities-in-q2-2026SecFeedde2026-08-26T12:02:51.666481+00:00Exploits and vulnerabilities in Q2 2026https://blog.schoergendorfer.com/post/choose-your-fighter-balancing-competing-requirements-to-select-models-for-your-aSecFeedde2026-08-26T12:02:51.581663+00:00Choose your fighter: Balancing competing requirements to select models for your AI SOChttps://blog.schoergendorfer.com/post/ai-models-flub-these-intelligence-tests-can-you-fare-any-betterSecFeedde2026-08-26T12:02:50.702030+00:00AI models flub these intelligence tests. Can you fare any better?https://blog.schoergendorfer.com/post/raised-on-aiSecFeedde2026-08-26T11:47:31.220314+00:00Raised on AIhttps://blog.schoergendorfer.com/post/the-missing-runtime-for-long-running-ai-agentsSecFeedde2026-08-26T11:47:30.438900+00:00The Missing Runtime for Long-Running AI Agentshttps://blog.schoergendorfer.com/post/automated-diagnosis-isnt-automated-understanding-what-postmortems-teach-us-aboutSecFeedde2026-08-26T11:47:30.401936+00:00Automated Diagnosis Isn’t Automated Understanding: What Postmortems Teach Us About Building Trustworthy Incident AIhttps://blog.schoergendorfer.com/post/ai-can-generate-your-infrastructure-can-your-cicd-pipeline-trust-itSecFeedde2026-08-26T11:32:52.752236+00:00AI Can Generate Your Infrastructure. Can Your CI/CD Pipeline Trust It?https://blog.schoergendorfer.com/post/your-ai-coding-budget-is-becoming-a-variable-cloud-billSecFeedde2026-08-26T11:17:34.323668+00:00Your AI Coding Budget Is Becoming a Variable Cloud Billhttps://blog.schoergendorfer.com/post/bill-gates-says-weve-passed-ais-danger-thresholds-now-whatSecFeedde2026-08-26T11:02:52.893030+00:00Bill Gates says we’ve passed AI’s danger thresholds. Now what?https://blog.schoergendorfer.com/post/amazon-gamelift-servers-enhanced-ddos-protection-now-availableSecFeedde2026-08-26T10:47:50.174644+00:00Amazon GameLift Servers enhanced DDoS Protection now availablehttps://blog.schoergendorfer.com/post/amazon-ec2-m8i-and-m8i-flex-instances-are-now-available-in-canada-west-calgary-rSecFeedde2026-08-26T10:47:50.142346+00:00Amazon EC2 M8i and M8i-flex instances are now available in Canada West (Calgary) regionhttps://blog.schoergendorfer.com/post/ibm-releases-granite-42-bringing-native-reasoning-and-agentic-rl-to-open-enterprSecFeedde2026-08-26T10:47:50.036500+00:00IBM Releases Granite 4.2: Bringing Native Reasoning and Agentic RL to Open Enterprise Modelshttps://blog.schoergendorfer.com/post/att-and-microsoft-scale-trillion-token-workloads-with-microsoft-foundry-and-amd-2SecFeedde2026-08-26T10:47:50.008951+00:00AT&T and Microsoft scale trillion-token workloads with Microsoft Foundry and AMDhttps://blog.schoergendorfer.com/post/the-economics-of-agent-optimization-from-pilots-to-measurable-returns-2SecFeedde2026-08-26T10:47:49.977492+00:00The Economics of Agent Optimization: From pilots to measurable returnshttps://blog.schoergendorfer.com/post/microsoft-named-a-leader-in-the-2026-gartner-magic-quadranttm-for-cloud-native-a-2SecFeedde2026-08-26T10:47:49.949705+00:00Microsoft named a Leader in the 2026 Gartner® Magic Quadrant™ for Cloud-Native Application Platformshttps://blog.schoergendorfer.com/post/the-patch-window-is-collapsing-why-security-needs-a-new-control-plane-2SecFeedde2026-08-26T10:47:49.910741+00:00The patch window is collapsing: Why security needs a new control planehttps://blog.schoergendorfer.com/post/constructcie-a-dataset-for-extracting-causal-information-from-construction-accid-3SecFeedde2026-08-26T10:47:49.824471+00:00ConstructCIE: A Dataset for Extracting Causal Information from Construction Accident Narrativeshttps://blog.schoergendorfer.com/post/rig-rope-relation-stratified-multimodal-attention-with-instance-local-rotary-geo-3SecFeedde2026-08-26T10:47:49.798757+00:00RIG-RoPE: Relation-Stratified Multimodal Attention with Instance-Local Rotary Geometry and Representation-Aware Traversal Coordinateshttps://blog.schoergendorfer.com/post/robustness-of-ir-models-to-collection-growth-2SecFeedde2026-08-26T10:47:49.773439+00:00Robustness of IR Models to Collection Growthhttps://blog.schoergendorfer.com/post/expectations-and-practices-around-ai-disclosure-in-cs-research-2SecFeedde2026-08-26T10:47:49.750149+00:00Expectations and Practices around AI Disclosure in CS Researchhttps://blog.schoergendorfer.com/post/cyberfactory-scaling-cyber-security-capabilities-with-instances-from-the-wild-2SecFeedde2026-08-26T10:47:49.722873+00:00CyberFactory: Scaling Cyber Security Capabilities with Instances from the Wildhttps://blog.schoergendorfer.com/post/a-comprehensive-analysis-of-arabic-natural-language-processing-research-trends-t-2SecFeedde2026-08-26T10:32:58.698217+00:00A Comprehensive Analysis of Arabic Natural Language Processing Research: Trends, Topic Evolution, and Research Gaps -- A Bibliometric and Topic-Based Studyhttps://blog.schoergendorfer.com/post/beyond-the-stability-exploration-dilemma-environmental-regularization-for-llm-po-2SecFeedde2026-08-26T10:32:58.671612+00:00Beyond the Stability-Exploration Dilemma: Environmental Regularization for LLM Policy Optimizationhttps://blog.schoergendorfer.com/post/longwof-bench-evaluating-evomap-genes-for-verifiable-long-workflow-tasks-2SecFeedde2026-08-26T10:32:58.644825+00:00LongWoF-Bench: Evaluating EvoMap Genes for Verifiable Long-Workflow Taskshttps://blog.schoergendorfer.com/post/counter-with-evidence-a-multi-agent-memory-efficient-reasoning-framework-for-hat-2SecFeedde2026-08-26T10:32:58.618873+00:00Counter with Evidence! A Multi-Agent Memory Efficient Reasoning Framework for Hate Category Informed Counterspeech Generationhttps://blog.schoergendorfer.com/post/most-of-the-llm-routing-gap-is-task-type-2SecFeedde2026-08-26T10:32:58.592673+00:00Most of the LLM Routing Gap Is Task Typehttps://blog.schoergendorfer.com/post/better-retrieval-worse-robustness-how-multi-hop-rag-amplifies-upstream-asr-errorSecFeedde2026-08-26T10:32:58.565152+00:00Better Retrieval, Worse Robustness: How Multi-hop RAG Amplifies Upstream ASR Errorshttps://blog.schoergendorfer.com/post/when-more-references-hurt-contamination-aware-dinov2-memory-banks-for-few-shot-sSecFeedde2026-08-26T10:32:58.539774+00:00When More References Hurt: Contamination-Aware DINOv2 Memory Banks for Few-Shot Steel Defect Detectionhttps://blog.schoergendorfer.com/post/autonomous-cyber-defense-real-time-attack-detection-and-mitigation-in-software-dSecFeedde2026-08-26T10:32:58.515332+00:00Autonomous Cyber Defense: Real-Time Attack Detection and Mitigation in Software-Defined Networks Using Machine Learninghttps://blog.schoergendorfer.com/post/gauss-hermite-quadrature-for-gaussian-mixture-entropy-with-an-action-space-hermiSecFeedde2026-08-26T10:32:58.489115+00:00Gauss--Hermite Quadrature for Gaussian-Mixture Entropy with an Action-Space Hermite Surrogatehttps://blog.schoergendorfer.com/post/if-it-walks-like-an-arbitrage-protocol-agnostic-detection-with-decidable-structu-2SecFeedde2026-08-26T10:32:58.466220+00:00If It Walks Like an Arbitrage: Protocol-Agnostic Detection with Decidable Structural Equivalencehttps://blog.schoergendorfer.com/post/representation-is-not-enough-body-localized-thermal-evidence-for-contactless-str-2SecFeedde2026-08-26T10:32:58.439721+00:00Representation Is Not Enough: Body-Localized Thermal Evidence for Contactless Stress and Craving Sensing in Opioid Use Disorderhttps://blog.schoergendorfer.com/post/gated-recurrent-transformers-expressive-depth-through-recurrent-modulation-in-trSecFeedde2026-08-26T10:32:58.387831+00:00Gated Recurrent Transformers: Expressive Depth through Recurrent Modulation in Transformershttps://blog.schoergendorfer.com/post/covariance-boosted-gaussian-processes-for-spatiotemporal-irregularities-2SecFeedde2026-08-26T10:32:58.356524+00:00Covariance-Boosted Gaussian Processes for Spatiotemporal Irregularitieshttps://blog.schoergendorfer.com/post/tool-making-and-self-evolving-llm-agents-in-low-latency-systems-2SecFeedde2026-08-26T10:32:58.329042+00:00Tool-Making and Self-Evolving LLM Agents in Low-Latency Systemshttps://blog.schoergendorfer.com/post/calibration-family-overfit-why-trusted-sabotage-monitors-dont-transfer-across-li-3SecFeedde2026-08-26T10:32:58.275083+00:00Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineageshttps://blog.schoergendorfer.com/post/geometric-bias-in-eigenspace-perturbation-under-random-heterogeneous-noiseSecFeedde2026-08-26T10:32:58.222780+00:00Geometric bias in eigenspace perturbation under random heterogeneous noisehttps://blog.schoergendorfer.com/post/from-local-geometry-to-global-pseudo-labeling-for-robust-positive-unlabeled-learSecFeedde2026-08-26T10:17:23.030178+00:00From Local Geometry to Global Pseudo Labeling for Robust Positive Unlabeled Learning under Covariate Shifthttps://blog.schoergendorfer.com/post/encrypted-neural-networks-without-overflowsSecFeedde2026-08-26T10:17:23.000610+00:00Encrypted Neural Networks without Overflowshttps://blog.schoergendorfer.com/post/the-theorems-of-dr-david-blackwell-and-their-contributions-to-artificial-intelliSecFeedde2026-08-26T10:17:22.950848+00:00The Theorems of Dr. David Blackwell and Their Contributions to Artificial Intelligencehttps://blog.schoergendorfer.com/post/naima-semantics-aware-rgb-guided-depth-super-resolutionSecFeedde2026-08-26T10:17:22.923158+00:00NAIMA: Semantics Aware RGB Guided Depth Super-Resolutionhttps://blog.schoergendorfer.com/post/bayes-with-no-shame-admissibility-geometries-of-predictive-inferenceSecFeedde2026-08-26T10:17:22.878190+00:00Bayes with No Shame: Admissibility Geometries of Predictive Inferencehttps://blog.schoergendorfer.com/post/do-physics-informed-neural-networks-pinns-need-to-be-deep-shallow-pinns-using-thSecFeedde2026-08-26T10:17:22.848900+00:00Do physics-informed neural networks (PINNs) need to be deep? Shallow PINNs using the Levenberg-Marquardt algorithmhttps://blog.schoergendorfer.com/post/mpib-a-benchmark-for-medical-prompt-injection-attacks-and-clinical-safety-in-llmSecFeedde2026-08-26T10:17:22.821856+00:00MPIB: A Benchmark for Medical Prompt Injection Attacks and Clinical Safety in LLMshttps://blog.schoergendorfer.com/post/e2hil-entropy-guided-sample-selection-for-efficient-real-world-human-in-the-loopSecFeedde2026-08-26T10:17:22.796684+00:00E2HiL: Entropy-Guided Sample Selection for Efficient Real-World Human-in-the-Loop Reinforcement Learninghttps://blog.schoergendorfer.com/post/reconquering-bell-sampling-on-qudits-stabilizer-learning-and-testing-quantum-pseSecFeedde2026-08-26T10:17:22.753111+00:00Reconquering Bell sampling on qudits: stabilizer learning and testing, quantum pseudorandomness bounds, and morehttps://blog.schoergendorfer.com/post/iwin-transformer-hierarchical-vision-transformer-using-interleaved-windowsSecFeedde2026-08-26T10:17:22.700814+00:00Iwin Transformer: Hierarchical Vision Transformer using Interleaved Windowshttps://blog.schoergendorfer.com/post/contextual-online-uncertainty-aware-preference-learning-for-human-feedbackSecFeedde2026-08-26T10:17:22.614481+00:00Contextual Online Uncertainty-Aware Preference Learning for Human Feedbackhttps://blog.schoergendorfer.com/post/two-sided-nearest-neighbors-an-adaptive-and-minimax-optimal-procedure-for-matrixSecFeedde2026-08-26T10:17:22.588584+00:00Two-Sided Nearest Neighbors: An adaptive and minimax optimal procedure for matrix completionhttps://blog.schoergendorfer.com/post/gnnbleed-inference-attacks-to-unveil-private-edges-in-graphs-with-realistic-acceSecFeedde2026-08-26T10:17:22.559215+00:00GNNBleed: Inference Attacks to Unveil Private Edges in Graphs with Realistic Access to GNN Modelshttps://blog.schoergendorfer.com/post/racr-mil-rank-aware-contextual-reasoning-for-weakly-supervised-grading-of-squamoSecFeedde2026-08-26T10:17:22.533393+00:00RACR-MIL: Rank-aware contextual reasoning for weakly supervised grading of squamous cell carcinoma using whole slide imageshttps://blog.schoergendorfer.com/post/nonconvex-nonconcave-min-max-optimization-with-a-small-maximization-domainSecFeedde2026-08-26T10:03:46.421081+00:00Nonconvex-Nonconcave Min-Max Optimization with a Small Maximization Domainhttps://blog.schoergendorfer.com/post/deep-feature-pyramid-convolutional-networks-with-in-place-activated-batch-normalSecFeedde2026-08-26T10:03:46.366745+00:00Deep Feature Pyramid Convolutional Networks with In-Place Activated Batch Normalization for Automated Skin Lesion Boundary Segmentationhttps://blog.schoergendorfer.com/post/spectrum-aware-bounds-on-invertibility-for-privacy-enhancing-instance-encodingSecFeedde2026-08-26T10:03:46.306795+00:00Spectrum-Aware Bounds on Invertibility for Privacy-Enhancing Instance Encodinghttps://blog.schoergendorfer.com/post/change-detection-in-probability-flow-ode-online-testing-in-diffusion-latent-spacSecFeedde2026-08-26T10:03:46.273138+00:00Change Detection in Probability Flow ODE: Online Testing in Diffusion Latent Spaceshttps://blog.schoergendorfer.com/post/blockwise-stabilized-adaptive-cubic-regularization-with-subsolvers-via-recurrencSecFeedde2026-08-26T10:03:46.244040+00:00Blockwise Stabilized Adaptive Cubic Regularization with Subsolvers via Recurrencehttps://blog.schoergendorfer.com/post/across-design-uncertainty-in-short-pricing-panels-inference-and-identificationSecFeedde2026-08-26T10:03:46.214786+00:00Across-Design Uncertainty in Short Pricing Panels: Inference and Identificationhttps://blog.schoergendorfer.com/post/metag-a-dataset-to-build-agentic-meta-reviewing-capabilities-2SecFeedde2026-08-26T10:03:46.179436+00:00Metag: A dataset to build agentic meta-reviewing capabilitieshttps://blog.schoergendorfer.com/post/multi-source-complex-network-reconstruction-via-wasserstein-distributionally-robSecFeedde2026-08-26T10:03:46.146000+00:00Multi-Source Complex Network Reconstruction via Wasserstein Distributionally Robust Optimization and Algorithm Unrollinghttps://blog.schoergendorfer.com/post/gear-generative-expansion-and-real-anchoring-for-two-stage-distillation-of-tabul-2SecFeedde2026-08-26T10:03:46.110354+00:00GEAR: Generative Expansion and Real Anchoring for Two-Stage Distillation of Tabular Foundation Modelshttps://blog.schoergendorfer.com/post/the-impact-of-temporal-context-length-and-encoding-strategies-on-self-supervised-2SecFeedde2026-08-26T10:03:46.075216+00:00The Impact of Temporal Context Length and Encoding Strategies on Self-Supervised ECG Representation Learninghttps://blog.schoergendorfer.com/post/an-efficient-minimax-optimal-algorithm-for-adversarial-m-set-banditsSecFeedde2026-08-26T10:03:46.040576+00:00An Efficient Minimax-Optimal Algorithm for Adversarial $m$-Set Banditshttps://blog.schoergendorfer.com/post/reproducible-evaluation-of-moe-expert-caching-replay-semantics-workload-contaminSecFeedde2026-08-26T10:03:46.006273+00:00Reproducible Evaluation of MoE Expert Caching: Replay Semantics, Workload Contamination, and Operating Regimeshttps://blog.schoergendorfer.com/post/when-may-a-model-replace-the-experiment-audits-licenses-and-the-price-of-trust-i-2SecFeedde2026-08-26T10:03:45.970292+00:00When May a Model Replace the Experiment? Audits, Licenses, and the Price of Trust in Surrogate-Driven Designhttps://blog.schoergendorfer.com/post/weak-to-strong-learning-in-decision-making-2SecFeedde2026-08-26T10:03:45.937452+00:00Weak-to-Strong Learning in Decision Makinghttps://blog.schoergendorfer.com/post/nonlinear-axiomatic-attribution-for-cooperative-games-2SecFeedde2026-08-26T10:03:45.901986+00:00Nonlinear Axiomatic Attribution for Cooperative Gameshttps://blog.schoergendorfer.com/post/application-of-machine-learning-to-monster-level-prediction-in-tabletop-rpg-game-3SecFeedde2026-08-26T10:03:45.869756+00:00Application of machine learning to monster level prediction in tabletop RPG game designhttps://blog.schoergendorfer.com/post/talk-text-attributed-graph-dataset-distillation-via-coupling-language-model-withSecFeedde2026-08-26T10:03:45.834201+00:00TaLK: Text-attributed Graph Dataset Distillation via Coupling Language Model with Graph-Aware Kernelhttps://blog.schoergendorfer.com/post/mortarbench-evaluating-mortgage-loan-origination-agentsSecFeedde2026-08-26T10:03:45.803648+00:00MortarBench: Evaluating Mortgage Loan Origination Agentshttps://blog.schoergendorfer.com/post/lightweight-adaptive-feature-composition-for-heterogeneous-downstream-adaptationSecFeedde2026-08-26T09:48:35.412011+00:00Lightweight Adaptive Feature Composition for Heterogeneous Downstream Adaptation of Wireless Foundation Modelshttps://blog.schoergendorfer.com/post/alpha-pfn-fast-entropy-search-via-in-context-learningSecFeedde2026-08-26T09:48:35.381955+00:00$\alpha$-PFN: Fast Entropy Search via In-Context Learninghttps://blog.schoergendorfer.com/post/model-based-learning-of-near-optimal-finite-window-policies-in-pomdpsSecFeedde2026-08-26T09:48:35.326405+00:00Model-Based Learning of Near-Optimal Finite-Window Policies in POMDPshttps://blog.schoergendorfer.com/post/a-bayesian-learning-approach-for-drone-coverage-network-a-case-study-on-cardiacSecFeedde2026-08-26T09:48:35.297761+00:00A Bayesian Learning Approach for Drone Coverage Network: A Case Study on Cardiac Arrest in Scotlandhttps://blog.schoergendorfer.com/post/breaking-the-tuning-barrier-zero-hyperparameters-yield-multi-corner-analysis-viaSecFeedde2026-08-26T09:48:35.262859+00:00Breaking the Tuning Barrier: Zero-Hyperparameters Yield Multi-Corner Analysis Via Learned Priorshttps://blog.schoergendorfer.com/post/how-to-achieve-the-intended-aim-of-deep-clustering-now-without-deep-learningSecFeedde2026-08-26T09:48:35.208641+00:00How to Achieve the Intended Aim of Deep Clustering Now, without Deep Learninghttps://blog.schoergendorfer.com/post/qimeng-chipv-rtl-exploiting-information-locality-for-ip-level-verilog-generationSecFeedde2026-08-26T09:48:35.180604+00:00QiMeng-ChipV-RTL: Exploiting Information Locality for IP-level Verilog Generationhttps://blog.schoergendorfer.com/post/wait-wait-wait-why-do-reasoning-models-loopSecFeedde2026-08-26T09:48:35.156303+00:00Wait, Wait, Wait... Why Do Reasoning Models Loop?https://blog.schoergendorfer.com/post/adaptive-prediction-theory-combining-offline-and-online-learningSecFeedde2026-08-26T09:48:35.129178+00:00Adaptive prediction theory combining offline and online learninghttps://blog.schoergendorfer.com/post/ltr-icd-a-ranking-aware-framework-for-automatic-icd-coding-2SecFeedde2026-08-26T09:48:35.096669+00:00LTR-ICD: A Ranking-Aware Framework for Automatic ICD Codinghttps://blog.schoergendorfer.com/post/sketchguard-scaling-byzantine-robust-decentralized-federated-learning-via-sketchSecFeedde2026-08-26T09:48:35.069517+00:00SketchGuard: Scaling Byzantine-Robust Decentralized Federated Learning via Sketch-Based Screeninghttps://blog.schoergendorfer.com/post/molga-molecular-graph-adaptation-with-pre-trained-2d-graph-encoderSecFeedde2026-08-26T09:48:35.039616+00:00MolGA: Molecular Graph Adaptation with Pre-trained 2D Graph Encoderhttps://blog.schoergendorfer.com/post/is-the-hard-label-cryptanalytic-model-extraction-really-polynomialSecFeedde2026-08-26T09:48:35.005203+00:00Is the Hard-Label Cryptanalytic Model Extraction Really Polynomial?https://blog.schoergendorfer.com/post/round-trip-reinforcement-learning-self-consistent-training-for-better-chemical-lSecFeedde2026-08-26T09:48:34.975273+00:00Round-trip Reinforcement Learning: Self-Consistent Training for Better Chemical LLMshttps://blog.schoergendorfer.com/post/massive-steps-massive-semantic-trajectories-for-understanding-poi-check-ins-dataSecFeedde2026-08-26T09:48:34.947506+00:00Massive-STEPS: Massive Semantic Trajectories for Understanding POI Check-ins -- Dataset and Benchmarkshttps://blog.schoergendorfer.com/post/stabilizing-temporal-difference-learning-via-implicit-stochastic-recursionSecFeedde2026-08-26T09:48:34.918795+00:00Stabilizing Temporal Difference Learning via Implicit Stochastic Recursionhttps://blog.schoergendorfer.com/post/tlxml-task-level-explanation-of-meta-learning-via-influence-functionsSecFeedde2026-08-26T09:48:34.891184+00:00TLXML: Task-Level Explanation of Meta-Learning via Influence Functionshttps://blog.schoergendorfer.com/post/qabba-error-guaranteed-symbolic-time-series-compression-via-integer-quantized-agSecFeedde2026-08-26T09:48:34.852859+00:00QABBA: Error-Guaranteed Symbolic Time-Series Compression via Integer-Quantized Aggregationhttps://blog.schoergendorfer.com/post/focal-calibration-loss-controlling-posterior-distortion-in-deep-neural-classifieSecFeedde2026-08-26T09:33:15.510242+00:00Focal Calibration Loss: Controlling Posterior Distortion in Deep Neural Classifiershttps://blog.schoergendorfer.com/post/quantum-maximum-entropy-inference-and-hamiltonian-learningSecFeedde2026-08-26T09:33:15.473716+00:00Quantum Maximum Entropy Inference and Hamiltonian Learninghttps://blog.schoergendorfer.com/post/polynomial-augmented-neural-networks-panns-with-weak-orthogonality-constraints-fSecFeedde2026-08-26T09:33:15.441614+00:00Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximationhttps://blog.schoergendorfer.com/post/adadagrad-adaptive-batch-size-schemes-for-adaptive-gradient-methodsSecFeedde2026-08-26T09:33:15.413592+00:00AdAdaGrad: Adaptive Batch Size Schemes for Adaptive Gradient Methodshttps://blog.schoergendorfer.com/post/what-fid-hides-detecting-ranking-and-diagnosing-deviations-in-generative-evaluatSecFeedde2026-08-26T09:33:15.358196+00:00What FID Hides: Detecting, Ranking, and Diagnosing Deviations in Generative Evaluationhttps://blog.schoergendorfer.com/post/parameterized-complexity-of-l_p-lipschitz-constants-for-input-convex-neural-netwSecFeedde2026-08-26T09:33:15.331790+00:00Parameterized Complexity of $L_p$-Lipschitz Constants for Input Convex Neural Networks and $L_p$-Norm Maximization over Zonotopeshttps://blog.schoergendorfer.com/post/mote-mixture-of-task-experts-for-multi-task-video-understandingSecFeedde2026-08-26T09:33:15.303742+00:00MoTE: Mixture of Task Experts for Multi-Task Video Understandinghttps://blog.schoergendorfer.com/post/weakly-supervised-seafloor-segmentation-for-seagrass-habitat-mapping-in-side-scaSecFeedde2026-08-26T09:33:15.273750+00:00Weakly Supervised Seafloor Segmentation for Seagrass Habitat Mapping in Side-Scan Sonar Imageryhttps://blog.schoergendorfer.com/post/when-similarity-is-interaction-driven-quantum-kernels-for-regime-sensitive-learnSecFeedde2026-08-26T09:33:15.245089+00:00When Similarity Is Interaction-Driven: Quantum Kernels for Regime-Sensitive Learninghttps://blog.schoergendorfer.com/post/morf-ast-calibrated-probabilistic-virtual-sensing-for-structural-monitoring-undeSecFeedde2026-08-26T09:33:15.218555+00:00MoRF-AST: Calibrated Probabilistic Virtual Sensing for Structural Monitoring under Changing Operating Conditionshttps://blog.schoergendorfer.com/post/satdl-jointly-optimizing-data-redistribution-and-training-for-satellite-based-diSecFeedde2026-08-26T09:33:15.169074+00:00SatDL: Jointly Optimizing Data Redistribution and Training for Satellite-Based Distributed Learninghttps://blog.schoergendorfer.com/post/neuralparker-a-reinforcement-learning-planner-for-irregular-parking-environmentsSecFeedde2026-08-26T09:33:15.141480+00:00NeuralParker: A Reinforcement Learning Planner for Irregular Parking Environmentshttps://blog.schoergendorfer.com/post/predictability-of-el-nino-from-delayed-observationsSecFeedde2026-08-26T09:33:15.088908+00:00Predictability of El Ni\~no from Delayed Observationshttps://blog.schoergendorfer.com/post/validation-of-hrv-studio-a-transparent-and-quality-control-aware-platform-for-heSecFeedde2026-08-26T09:33:15.040392+00:00Validation of HRV Studio: A Transparent and Quality-Control-Aware Platform for Heart Rate Variability Analysishttps://blog.schoergendorfer.com/post/a-theory-of-finite-noise-optima-and-generalization-in-quantum-machine-learningSecFeedde2026-08-26T09:33:15.006225+00:00A Theory of Finite-Noise Optima and Generalization in Quantum Machine Learninghttps://blog.schoergendorfer.com/post/a-heterogeneous-mixture-of-experts-framework-for-interpretable-machine-learningSecFeedde2026-08-26T09:33:14.966401+00:00A Heterogeneous Mixture of Experts Framework for Interpretable Machine Learninghttps://blog.schoergendorfer.com/post/decoupling-candidate-dual-agn-from-chance-superpositions-in-the-gothic-survey-viSecFeedde2026-08-26T09:17:27.774755+00:00Decoupling candidate dual AGN from chance superpositions in the GOTHIC survey via a deep-learning frameworkhttps://blog.schoergendorfer.com/post/anatomy-of-a-scam-call-what-10000-real-scam-and-spam-calls-reveal-about-how-phonSecFeedde2026-08-26T09:17:27.744618+00:00Anatomy of a Scam Call: What 10,000 real scam and spam calls reveal about how phone scammers operatehttps://blog.schoergendorfer.com/post/qshap-fast-shapley-decomposition-of-r2-for-gradient-boosted-treesSecFeedde2026-08-26T09:17:27.715402+00:00qshap: Fast Shapley Decomposition of $R^2$ for Gradient-Boosted Treeshttps://blog.schoergendorfer.com/post/retrievalformer-a-dual-encoder-transformer-for-efficient-approximate-nearest-neiSecFeedde2026-08-26T09:17:27.662310+00:00RetrievalFormer: A Dual-Encoder Transformer for Efficient Approximate Nearest Neighbor Retrieval and Cold-Item Recommendationhttps://blog.schoergendorfer.com/post/generalization-memorization-and-overfitting-for-diffusion-models-trained-in-theSecFeedde2026-08-26T09:17:27.631521+00:00Generalization, memorization, and overfitting for diffusion models trained in the lazy high-dimensional regimehttps://blog.schoergendorfer.com/post/dimensionless-controls-of-plasticity-under-alternating-tasks-from-evolutionary-bSecFeedde2026-08-26T09:17:27.597543+00:00Dimensionless Controls of Plasticity Under Alternating Tasks: From Evolutionary Biology to Continual Learninghttps://blog.schoergendorfer.com/post/spatiotemporal-distillation-via-recurrent-bottlenecks-for-aortic-trackingSecFeedde2026-08-26T09:17:27.563774+00:00Spatiotemporal Distillation via Recurrent Bottlenecks for Aortic Trackinghttps://blog.schoergendorfer.com/post/differential-learning-for-robust-prediction-of-thermal-stability-with-applicatioSecFeedde2026-08-26T09:17:27.528687+00:00Differential Learning for Robust Prediction of Thermal Stability with Application to Energetic Materialshttps://blog.schoergendorfer.com/post/pipeline-native-transformers-co-designing-model-architecture-and-cpu-inference-fSecFeedde2026-08-26T09:17:27.499748+00:00Pipeline-Native Transformers: Co-Designing Model Architecture and CPU Inference for Bandwidth-Efficient Autoregressive Decodehttps://blog.schoergendorfer.com/post/learning-to-act-while-waiting-rl-finetuning-of-generalist-robot-policies-under-iSecFeedde2026-08-26T09:17:27.471155+00:00Learning to Act While Waiting: RL Finetuning of Generalist Robot Policies Under Inference Latencyhttps://blog.schoergendorfer.com/post/mitigating-exploration-bias-in-rl-for-multi-instruction-followingSecFeedde2026-08-26T09:17:27.439552+00:00Mitigating Exploration Bias in RL for Multi-Instruction Followinghttps://blog.schoergendorfer.com/post/accelerating-the-adoption-of-residential-solar-power-systems-policy-analysis-usiSecFeedde2026-08-26T09:17:27.409850+00:00Accelerating the Adoption of Residential Solar Power Systems: Policy Analysis using a Dynamic Structural Modelhttps://blog.schoergendorfer.com/post/s-matrix-informed-neural-networks-for-amplitude-analysisSecFeedde2026-08-26T09:17:27.352860+00:00S-matrix informed neural networks for amplitude analysishttps://blog.schoergendorfer.com/post/a-hybrid-two-stage-machine-learning-pipeline-for-fault-detection-and-classificatSecFeedde2026-08-26T09:17:27.321785+00:00A Hybrid Two-Stage Machine Learning Pipeline for Fault Detection and Classification in Power Transmission Systemshttps://blog.schoergendorfer.com/post/scaling-reinforcement-learning-for-diffusion-models-via-velocity-matchingSecFeedde2026-08-26T09:17:27.289449+00:00Scaling Reinforcement Learning for Diffusion Models via Velocity Matchinghttps://blog.schoergendorfer.com/post/replicable-conformal-predictionSecFeedde2026-08-26T09:17:27.224434+00:00Replicable Conformal Predictionhttps://blog.schoergendorfer.com/post/the-blending-ratio-is-not-where-the-performance-is-diagnosing-prototype-blendingSecFeedde2026-08-26T09:17:27.191197+00:00The Blending Ratio Is Not Where the Performance Is: Diagnosing Prototype Blending for Few-Shot Adaptation of Vision-Language Modelshttps://blog.schoergendorfer.com/post/statetune-transforming-llm-assisted-eda-flow-tuning-into-a-stateful-closed-loopSecFeedde2026-08-26T09:02:47.374414+00:00StateTune: Transforming LLM-Assisted EDA Flow Tuning into a Stateful, Closed-Loop Processhttps://blog.schoergendorfer.com/post/transformer-accelerator-tfa-a-macro-op-int8-hardware-chip-for-transformer-infereSecFeedde2026-08-26T09:02:47.339931+00:00Transformer Accelerator (TFA): A Macro-Op INT8 Hardware Chip for Transformer Inference and Machine Translationhttps://blog.schoergendorfer.com/post/infodpp-pac-principled-patch-selection-for-whole-slide-image-analysisSecFeedde2026-08-26T09:02:47.311921+00:00InfoDPP-PAC: Principled Patch Selection for Whole Slide Image Analysishttps://blog.schoergendorfer.com/post/did-it-in-87-minutes-a-label-free-softmax-to-linear-adaptation-of-vision-transfoSecFeedde2026-08-26T09:02:47.279904+00:00DiD It in 87 Minutes: A Label-Free Softmax-to-Linear Adaptation of Vision Transformers for Object Detectionhttps://blog.schoergendorfer.com/post/finite-sample-metric-non-collapse-for-geometrically-supervised-latent-world-modeSecFeedde2026-08-26T09:02:47.248656+00:00Finite-Sample Metric Non-Collapse for Geometrically Supervised Latent World Models in Controlhttps://blog.schoergendorfer.com/post/symbolic-classification-enabled-lhc-limits-online-bsm-global-fitsSecFeedde2026-08-26T09:02:47.217724+00:00Symbolic Classification-Enabled LHC Limits Online BSM Global Fitshttps://blog.schoergendorfer.com/post/improving-cross-problem-vehicle-routing-with-locally-augmented-preferences-and-rSecFeedde2026-08-26T09:02:47.184454+00:00Improving Cross-Problem Vehicle Routing with Locally Augmented Preferences and Representation Disentanglementhttps://blog.schoergendorfer.com/post/bellman-calibration-for-marginalized-importance-weighting-in-offline-reinforcemeSecFeedde2026-08-26T09:02:47.150864+00:00Bellman Calibration for Marginalized Importance Weighting in Offline Reinforcement Learninghttps://blog.schoergendorfer.com/post/effective-learning-rate-governs-loss-dynamics-in-language-model-pretrainingSecFeedde2026-08-26T09:02:47.066450+00:00Effective Learning Rate Governs Loss Dynamics in Language Model Pretraininghttps://blog.schoergendorfer.com/post/mdte-minority-aware-diffusion-over-temporal-edge-events-for-imbalanced-node-clasSecFeedde2026-08-26T09:02:47.034103+00:00MDTE: Minority-Aware Diffusion over Temporal Edge Events for Imbalanced Node Classificationhttps://blog.schoergendorfer.com/post/lion-a-clifford-neural-paradigm-for-multimodal-attributed-graph-learningSecFeedde2026-08-26T09:02:46.999175+00:00LION: A Clifford Neural Paradigm for Multimodal-Attributed Graph Learninghttps://blog.schoergendorfer.com/post/textdnn2-doubly-non-negative-relaxations-for-deep-neural-networksSecFeedde2026-08-26T09:02:46.968562+00:00$(\text{DNN})^2$: Doubly Non-Negative Relaxations for Deep Neural Networkshttps://blog.schoergendorfer.com/post/optimal-alternating-regret-for-online-learning-and-gamesSecFeedde2026-08-26T09:02:46.934538+00:00Optimal Alternating Regret for Online Learning and Gameshttps://blog.schoergendorfer.com/post/parameter-level-attribution-of-symmetry-in-trained-networks-though-parameter-wisSecFeedde2026-08-26T09:02:46.900982+00:00Parameter-Level Attribution of Symmetry in Trained Networks Though Parameter-Wise Functional Sensitivityhttps://blog.schoergendorfer.com/post/a-multimodal-foundation-model-for-longitudinal-patient-representation-and-scalabSecFeedde2026-08-26T09:02:46.841665+00:00A Multimodal Foundation Model for Longitudinal Patient Representation and Scalable Insight Generation in Oncologyhttps://blog.schoergendorfer.com/post/data-leakage-inflates-generalizability-of-power-outage-prediction-modelsSecFeedde2026-08-26T09:02:46.813224+00:00Data Leakage Inflates Generalizability of Power Outage Prediction Modelshttps://blog.schoergendorfer.com/post/conditional-graphganfed-optimizing-graph-structured-molecule-generation-in-federSecFeedde2026-08-26T08:47:29.361134+00:00Conditional GraphGANFed: Optimizing Graph-Structured Molecule Generation in Federated Generative Adversarial Networkshttps://blog.schoergendorfer.com/post/delayed-optimizer-state-transport-shapes-short-horizon-training-decisionsSecFeedde2026-08-26T08:47:29.337663+00:00Delayed Optimizer-State Transport Shapes Short-Horizon Training Decisionshttps://blog.schoergendorfer.com/post/iapo-influence-aware-policy-optimization-for-credit-assignment-in-multi-turn-serSecFeedde2026-08-26T08:47:29.310711+00:00IAPO: Influence-Aware Policy Optimization for Credit Assignment in Multi-Turn Service Agentshttps://blog.schoergendorfer.com/post/seismamba-low-latency-single-station-seismic-magnitude-estimation-for-spatiallySecFeedde2026-08-26T08:47:29.284316+00:00SeisMamba: Low-Latency Single-Station Seismic Magnitude Estimation for Spatially Distributed Earthquake Early Warninghttps://blog.schoergendorfer.com/post/persistent-cross-entropySecFeedde2026-08-26T08:47:29.260983+00:00Persistent Cross Entropyhttps://blog.schoergendorfer.com/post/from-numerical-simulators-of-pdes-to-neural-emulators-and-backSecFeedde2026-08-26T08:47:29.238448+00:00From Numerical Simulators of PDEs to Neural Emulators and Backhttps://blog.schoergendorfer.com/post/it-depends-incorporating-correlations-for-joint-aleatoric-and-epistemic-uncertaiSecFeedde2026-08-26T08:47:29.209650+00:00It depends: Incorporating correlations for joint aleatoric and epistemic uncertainties of high-dimensional output spaceshttps://blog.schoergendorfer.com/post/where-entropy-is-measured-matters-policy-geometry-in-bounded-continuous-controlSecFeedde2026-08-26T08:47:29.181531+00:00Where Entropy Is Measured Matters: Policy Geometry in Bounded Continuous-Control PPOhttps://blog.schoergendorfer.com/post/warpsac-towards-the-pinnacle-of-scalable-off-policy-rl-by-rethinking-explorationSecFeedde2026-08-26T08:47:29.153669+00:00WarpSAC: Towards the Pinnacle of Scalable Off-policy RL by Rethinking Exploration and Exploitationhttps://blog.schoergendorfer.com/post/joint-distribution-alignment-for-universal-domain-adaptationSecFeedde2026-08-26T08:47:29.127399+00:00Joint Distribution Alignment for Universal Domain Adaptationhttps://blog.schoergendorfer.com/post/when-does-self-supervised-pretraining-help-tabular-models-a-study-of-label-scarcSecFeedde2026-08-26T08:47:29.102250+00:00When Does Self-Supervised Pretraining Help Tabular Models? A Study of Label Scarcity and Missing Datahttps://blog.schoergendorfer.com/post/a-structural-fhmm-for-interpretable-disease-trajectories-in-t2dmSecFeedde2026-08-26T08:47:29.076203+00:00A Structural FHMM for Interpretable Disease Trajectories in T2DMhttps://blog.schoergendorfer.com/post/causal-analysis-for-time-series-foundation-modelsSecFeedde2026-08-26T08:47:29.050551+00:00Causal Analysis for Time Series Foundation Modelshttps://blog.schoergendorfer.com/post/a-data-dependent-early-stopping-rule-using-rademacher-complexity-with-l1-normSecFeedde2026-08-26T08:47:29.023812+00:00A Data-dependent Early Stopping Rule using Rademacher Complexity with L1-normhttps://blog.schoergendorfer.com/post/prq-kmeans-projection-residual-quantization-for-semantic-id-tokenizationSecFeedde2026-08-26T08:47:28.993094+00:00PRQ-KMeans: Projection Residual Quantization for Semantic ID Tokenizationhttps://blog.schoergendorfer.com/post/from-relaxed-indexability-to-exact-indexability-a-t-step-approach-for-partiallySecFeedde2026-08-26T08:47:28.945518+00:00From Relaxed Indexability to Exact Indexability: A $t$-Step Approach for Partially Observable Restless Banditshttps://blog.schoergendorfer.com/post/robust-data-collection-policy-learning-for-low-variance-online-policy-evaluationSecFeedde2026-08-26T08:47:28.916139+00:00Robust Data-Collection Policy Learning for Low-Variance Online Policy Evaluationhttps://blog.schoergendorfer.com/post/steering-recurrent-reasoners-at-inference-time-with-readout-feedbackSecFeedde2026-08-26T08:47:28.886715+00:00Steering Recurrent Reasoners at Inference Time with Readout Feedbackhttps://blog.schoergendorfer.com/post/a-mesh-free-multiresolution-deep-energy-method-with-phase-field-modeling-of-britSecFeedde2026-08-26T08:47:28.855999+00:00A mesh-free multiresolution deep energy method with phase-field modeling of brittle fracturehttps://blog.schoergendorfer.com/post/the-sharp-tail-of-uniform-stabilitySecFeedde2026-08-26T08:47:28.815857+00:00The Sharp Tail of Uniform Stabilityhttps://blog.schoergendorfer.com/post/a-feature-major-codebook-for-memory-efficient-sparse-binary-self-organizing-mapsSecFeedde2026-08-26T08:32:57.678879+00:00A Feature-Major Codebook for Memory-Efficient Sparse-Binary Self-Organizing Maps: Scaling a MEDLINE Atlas to 1.05 Million Neurons on a Single Consumer GPUhttps://blog.schoergendorfer.com/post/physicsbench-a-unified-leaderboard-for-generative-and-predictive-models-in-enginSecFeedde2026-08-26T08:32:57.638575+00:00PhysicsBench: A Unified Leaderboard for Generative and Predictive Models in Engineering Design and Simulationhttps://blog.schoergendorfer.com/post/alphabet-a-laplace-pole-history-aggregator-with-banked-exponential-transportSecFeedde2026-08-26T08:32:57.600396+00:00ALPHABET: A Laplace-Pole History Aggregator with Banked Exponential Transporthttps://blog.schoergendorfer.com/post/physics-integrated-operator-learning-via-gaussian-splatting-representationsSecFeedde2026-08-26T08:32:57.566088+00:00Physics-Integrated Operator Learning via Gaussian Splatting Representationshttps://blog.schoergendorfer.com/post/xp-jepa-cross-predictive-physics-grounding-for-forecastable-latent-dynamicsSecFeedde2026-08-26T08:32:57.524805+00:00XP-JEPA: Cross-Predictive Physics Grounding for Forecastable Latent Dynamicshttps://blog.schoergendorfer.com/post/pinsieve-production-selective-vlm-serving-and-a-governed-memory-flywheel-for-entSecFeedde2026-08-26T08:32:57.489820+00:00PinSieve: Production Selective VLM Serving and a Governed Memory Flywheel for Enterprise Content-Quality Triagehttps://blog.schoergendorfer.com/post/low-latency-activation-regularized-sparse-neural-operators-with-distillation-assSecFeedde2026-08-26T08:32:57.424944+00:00Low-Latency Activation-Regularized Sparse Neural Operators with Distillation Assistance Towards Real-Time Edge-Deployable Virtual Sensinghttps://blog.schoergendorfer.com/post/codrift-compositional-drifting-for-offline-reinforcement-learningSecFeedde2026-08-26T08:32:57.386838+00:00CoDrift: Compositional Drifting for Offline Reinforcement Learninghttps://blog.schoergendorfer.com/post/the-loss-floor-of-denoising-score-matching-fisher-geometry-from-schrodinger-bridSecFeedde2026-08-26T08:32:57.282507+00:00The Loss Floor of Denoising Score Matching: Fisher Geometry from Schr\"odinger Bridgeshttps://blog.schoergendorfer.com/post/partial-optimal-transport-on-the-circle-for-all-transported-masses-in-on-log-nSecFeedde2026-08-26T08:32:57.241931+00:00Partial Optimal Transport on the Circle for All Transported Masses in O(N log N)https://blog.schoergendorfer.com/post/every-layer-counts-an-exponential-l_2-depth-hierarchy-for-relu-networksSecFeedde2026-08-26T08:32:57.204102+00:00Every Layer Counts: An Exponential $L_2$ Depth Hierarchy for ReLU Networkshttps://blog.schoergendorfer.com/post/uhi-bench-benchmarking-dual-source-urban-heat-island-modeling-across-cities-in-dSecFeedde2026-08-26T08:32:57.172871+00:00UHI-Bench: Benchmarking Dual-Source Urban Heat Island Modeling Across Cities in Diverse Climate Regimeshttps://blog.schoergendorfer.com/post/flowneg-gflownet-guided-diverse-hard-negative-sampling-for-knowledge-graph-embedSecFeedde2026-08-26T08:32:57.136880+00:00FlowNeg: GFlowNet-Guided Diverse Hard Negative Sampling for Knowledge Graph Embeddinghttps://blog.schoergendorfer.com/post/puzzlekv-page-wise-low-rank-decomposition-for-kv-cache-compressionSecFeedde2026-08-26T08:32:57.102667+00:00PuzzleKV: Page-Wise Low-Rank Decomposition for KV Cache Compressionhttps://blog.schoergendorfer.com/post/generating-intervention-hypotheses-using-explainable-explanations-on-graphs-g2iSecFeedde2026-08-26T08:32:57.071382+00:00Generating Intervention Hypotheses using Explainable Explanations on Graphs: G2I, a Two-Stage Greedy Frameworkhttps://blog.schoergendorfer.com/post/aqlora-a-zero-search-recipe-for-fast-quantized-lora-fine-tuningSecFeedde2026-08-26T08:32:57.034166+00:00AQLoRA: A Zero-Search Recipe for Fast Quantized LoRA Fine-Tuninghttps://blog.schoergendorfer.com/post/a-theory-of-speciation-in-generative-diffusion-models-on-compact-riemannian-maniSecFeedde2026-08-26T08:32:56.998179+00:00A Theory of Speciation in Generative Diffusion Models on Compact Riemannian Manifoldshttps://blog.schoergendorfer.com/post/mixture-of-channel-experts-static-sparse-supports-with-input-adaptive-mixing-forSecFeedde2026-08-26T08:17:37.633417+00:00Mixture of Channel Experts: Static Sparse Supports with Input-Adaptive Mixing for Pointwise Projectionshttps://blog.schoergendorfer.com/post/gap-prompt-gated-adaptive-prompting-for-efficient-continual-learningSecFeedde2026-08-26T08:17:37.604502+00:00GAP-Prompt: Gated Adaptive Prompting for Efficient Continual Learninghttps://blog.schoergendorfer.com/post/tight-majorizations-and-convergence-rates-of-nuclear-norm-minimization-irlsSecFeedde2026-08-26T08:17:37.576833+00:00Tight Majorizations and Convergence Rates of Nuclear Norm Minimization IRLShttps://blog.schoergendorfer.com/post/calibration-preserving-pruning-compression-as-a-reliability-contractSecFeedde2026-08-26T08:17:37.543377+00:00Calibration-Preserving Pruning: Compression as a Reliability Contracthttps://blog.schoergendorfer.com/post/response-renormalization-for-critical-deep-equilibrium-modelsSecFeedde2026-08-26T08:17:37.517716+00:00Response Renormalization for Critical Deep Equilibrium Modelshttps://blog.schoergendorfer.com/post/renormalization-group-flow-matching-for-scalable-local-generative-modelingSecFeedde2026-08-26T08:17:37.489243+00:00Renormalization Group Flow Matching for Scalable Local Generative Modelinghttps://blog.schoergendorfer.com/post/data-predictability-shapes-weibull-weight-scale-growth-in-transformer-trainingSecFeedde2026-08-26T08:17:37.461028+00:00Data Predictability Shapes Weibull Weight-Scale Growth in Transformer Traininghttps://blog.schoergendorfer.com/post/equivariant-cellular-sheaves-for-molecular-electronic-structure-bridging-sheaf-cSecFeedde2026-08-26T08:17:37.431764+00:00Equivariant Cellular Sheaves for Molecular Electronic Structure: Bridging Sheaf Cohomology and E(3)-Equivariant Hamiltonian Learninghttps://blog.schoergendorfer.com/post/best-practice-critic-optimizationSecFeedde2026-08-26T08:17:37.402340+00:00Best Practice Critic Optimizationhttps://blog.schoergendorfer.com/post/whats-the-catch-evaluating-temporal-consistency-in-vision-language-models-2SecFeedde2026-08-26T08:17:37.373353+00:00What's the Catch? Evaluating Temporal Consistency in Vision-Language Modelshttps://blog.schoergendorfer.com/post/cross-domain-multi-task-data-to-text-generation-without-in-domain-training-data-2SecFeedde2026-08-26T08:17:37.343804+00:00Cross-Domain, Multi-Task Data-to-Text Generation without In-Domain Training Datahttps://blog.schoergendorfer.com/post/thinking-beyond-videos-unifying-video-reasoning-and-deep-research-for-open-world-2SecFeedde2026-08-26T08:17:37.315244+00:00Thinking Beyond Videos: Unifying Video Reasoning and Deep Research for Open-World Video Agentshttps://blog.schoergendorfer.com/post/how-much-regularization-survives-averaging-update-masking-in-federated-learning-2SecFeedde2026-08-26T08:17:37.292641+00:00How Much Regularization Survives Averaging? Update Masking in Federated Learninghttps://blog.schoergendorfer.com/post/molecular-llm-agents-from-architectural-design-to-scientific-autonomy-2SecFeedde2026-08-26T08:17:37.269822+00:00Molecular LLM Agents: From Architectural Design to Scientific Autonomyhttps://blog.schoergendorfer.com/post/sane-state-anomaly-neutralization-for-stable-extreme-context-delta-rule-models-2SecFeedde2026-08-26T08:17:37.200840+00:00SANE: State Anomaly Neutralization for Stable Extreme-Context Delta-Rule Modelshttps://blog.schoergendorfer.com/post/inferring-action-from-future-latent-state-for-robotic-manipulation-2SecFeedde2026-08-26T08:17:37.178155+00:00Inferring Action from Future Latent State for Robotic Manipulationhttps://blog.schoergendorfer.com/post/training-a-knowledge-base-supervised-structure-learning-for-agent-curated-docume-2SecFeedde2026-08-26T08:17:37.153291+00:00Training a Knowledge Base: Supervised Structure Learning for Agent-Curated Document Storeshttps://blog.schoergendorfer.com/post/is-visual-prompting-all-you-need-studying-vlm-spatial-reasoning-under-progressiv-2SecFeedde2026-08-26T08:17:37.125821+00:00Is Visual Prompting All You Need? Studying VLM Spatial Reasoning under Progressive Visual Scaffoldshttps://blog.schoergendorfer.com/post/civa-critic-induced-value-subspace-attacks-on-visual-world-model-agents-2SecFeedde2026-08-26T08:02:53.407557+00:00CIVA: Critic-Induced Value-Subspace Attacks on Visual World-Model Agentshttps://blog.schoergendorfer.com/post/scaling-muon-for-diffusion-transformers-2SecFeedde2026-08-26T08:02:53.378395+00:00Scaling Muon for Diffusion Transformershttps://blog.schoergendorfer.com/post/exploratwin-a-non-profit-research-platform-for-digital-twin-simulations-2SecFeedde2026-08-26T08:02:53.322751+00:00ExploraTwin, a Non-Profit Research Platform for Digital Twin Simulationshttps://blog.schoergendorfer.com/post/credit-without-ground-truth-auditing-step-level-credit-assignment-in-llm-agents-2SecFeedde2026-08-26T08:02:53.297081+00:00Credit Without Ground Truth: Auditing Step-Level Credit Assignment in LLM Agents Against Executed Replayhttps://blog.schoergendorfer.com/post/formal-verification-of-romanovs-triplet-logic-a-verified-filter-for-sliding-wind-3SecFeedde2026-08-26T08:02:53.269085+00:00Formal Verification of Romanov's Triplet Logic: A Verified Filter for Sliding-window 3-CNF with Application to Structured Formulashttps://blog.schoergendorfer.com/post/from-corpora-to-co-evolving-capabilities-capability-centric-data-design-for-gene-2SecFeedde2026-08-26T08:02:53.241251+00:00From Corpora to Co-Evolving Capabilities: Capability-Centric Data Design for Generalist Image Generationhttps://blog.schoergendorfer.com/post/maple-moe-adaptive-plug-and-play-layer-wise-expert-allocation-2SecFeedde2026-08-26T08:02:53.214114+00:00MAPLE: MoE Adaptive Plug-and-play Layer-wise Expert allocationhttps://blog.schoergendorfer.com/post/er-kans-efficient-and-robust-kolmogorov-arnold-networks-for-data-scarce-scientif-2SecFeedde2026-08-26T08:02:53.180419+00:00ER-KANs: Efficient and Robust Kolmogorov-Arnold Networks for Data-Scarce Scientific Machine Learninghttps://blog.schoergendorfer.com/post/epistemic-transfer-in-ai-assisted-verification-a-framework-and-evaluation-protoc-4SecFeedde2026-08-26T08:02:53.145877+00:00Epistemic Transfer in AI-Assisted Verification: A Framework and Evaluation Protocolhttps://blog.schoergendorfer.com/post/aerodpo-unleashing-lightweight-uav-navigation-with-high-fidelity-perception-and-2SecFeedde2026-08-26T08:02:53.111617+00:00AeroDPO: Unleashing Lightweight UAV Navigation with High-Fidelity Perception and Automated Preference Optimizationhttps://blog.schoergendorfer.com/post/audio-to-score-transcription-using-pre-trained-features-data-augmentation-and-th-3SecFeedde2026-08-26T08:02:53.079881+00:00Audio-to-Score Transcription using Pre-trained Features, Data Augmentation, and the New SheetSage-A2S Datasethttps://blog.schoergendorfer.com/post/mosaic-masked-outsourcing-of-secure-ai-computations-2SecFeedde2026-08-26T08:02:53.024704+00:00MOSAIC: Masked Outsourcing of Secure AI Computationshttps://blog.schoergendorfer.com/post/locks-page-local-compact-key-summaries-for-efficient-long-context-decoding-2SecFeedde2026-08-26T08:02:52.994074+00:00LOCKS: Page-Local Compact Key Summaries for Efficient Long-Context Decodinghttps://blog.schoergendorfer.com/post/graphvid-interactive-graph-controllable-video-generation-2SecFeedde2026-08-26T08:02:52.961824+00:00GraphVid: Interactive Graph-Controllable Video Generationhttps://blog.schoergendorfer.com/post/evipathbench-benchmarking-evidence-acquisition-and-reasoning-in-vision-languageSecFeedde2026-08-26T08:02:52.930652+00:00EviPathBench: Benchmarking Evidence Acquisition and Reasoning in Vision-Language Models for Whole-Slide Pathologyhttps://blog.schoergendorfer.com/post/the-cafe-in-amsterdam-when-the-incumbent-becomes-the-oracle-3SecFeedde2026-08-26T08:02:52.874390+00:00The Caf\'e in Amsterdam: When the Incumbent Becomes the Oraclehttps://blog.schoergendorfer.com/post/eluna-an-agentic-llm-system-for-automating-warehouse-operations-with-reasoning-a-2SecFeedde2026-08-26T08:02:52.838657+00:00Eluna: An Agentic LLM System for Automating Warehouse Operations with Reasoning and Task Executionhttps://blog.schoergendorfer.com/post/a-unified-algebraic-framework-for-classification-performance-evaluation-2SecFeedde2026-08-26T07:47:26.623421+00:00A Unified Algebraic Framework for Classification Performance Evaluationhttps://blog.schoergendorfer.com/post/optimizing-expert-designed-crystal-graph-networks-for-band-gap-prediction-with-a-2SecFeedde2026-08-26T07:47:26.595334+00:00Optimizing Expert-Designed Crystal Graph Networks for Band-Gap Prediction with an Autonomous LLM Research Loophttps://blog.schoergendorfer.com/post/co-occurring-associated-retained-concepts-in-diffusion-unlearningSecFeedde2026-08-26T07:47:26.569189+00:00Co-occurring Associated REtained concepts in Diffusion Unlearninghttps://blog.schoergendorfer.com/post/rarm-confidence-gated-progress-reward-modeling-for-rl-in-manipulation-2SecFeedde2026-08-26T07:47:26.540973+00:00RARM: Confidence-Gated Progress Reward Modeling for RL in Manipulationhttps://blog.schoergendorfer.com/post/tw-legalbench-measuring-taiwanese-legal-understandingSecFeedde2026-08-26T07:47:26.512574+00:00TW-LegalBench: Measuring Taiwanese Legal Understandinghttps://blog.schoergendorfer.com/post/when-can-one-neuron-fix-repetition-loops-in-llmsSecFeedde2026-08-26T07:47:26.485300+00:00When Can One Neuron Fix Repetition Loops in LLMs?https://blog.schoergendorfer.com/post/a-circuit-not-the-circuit-non-unique-causal-localisation-of-the-mamba-2-state-siSecFeedde2026-08-26T07:47:26.438094+00:00A Circuit, Not The Circuit: Non-Unique Causal Localisation of the Mamba-2 State Sinkhttps://blog.schoergendorfer.com/post/skill-conditioned-gated-self-distillation-for-llm-reasoningSecFeedde2026-08-26T07:47:26.412645+00:00Skill-Conditioned Gated Self-Distillation for LLM Reasoninghttps://blog.schoergendorfer.com/post/tournament-grpo-group-wise-tournament-rewards-for-reinforcement-learning-in-openSecFeedde2026-08-26T07:47:26.382188+00:00Tournament-GRPO: Group-Wise Tournament Rewards for Reinforcement Learning in Open-Ended Long-Form Generationhttps://blog.schoergendorfer.com/post/forceflow-learning-to-feel-and-act-via-contact-driven-flow-matchingSecFeedde2026-08-26T07:47:26.354329+00:00ForceFlow: Learning to Feel and Act via Contact-Driven Flow Matchinghttps://blog.schoergendorfer.com/post/coworld-vla-thinking-in-a-multi-expert-world-model-for-autonomous-drivingSecFeedde2026-08-26T07:47:26.326450+00:00CoWorld-VLA: Thinking in a Multi-Expert World Model for Autonomous Drivinghttps://blog.schoergendorfer.com/post/outlier-robust-diffusion-solvers-for-inverse-problemsSecFeedde2026-08-26T07:47:26.300257+00:00Outlier-Robust Diffusion Solvers for Inverse Problemshttps://blog.schoergendorfer.com/post/superintelligent-retrieval-agent-the-next-frontier-of-agentic-retrievalSecFeedde2026-08-26T07:47:26.272402+00:00Superintelligent Retrieval Agent: The Next Frontier of Agentic Retrievalhttps://blog.schoergendorfer.com/post/ra-cmf-region-adaptive-conditional-meanflow-for-ct-image-reconstructionSecFeedde2026-08-26T07:47:26.213134+00:00RA-CMF: Region-Adaptive Conditional MeanFlow for CT Image Reconstructionhttps://blog.schoergendorfer.com/post/test-time-adaptation-for-eeg-foundation-models-a-systematic-study-under-real-wor-2SecFeedde2026-08-26T07:47:26.186127+00:00Test-Time Adaptation for EEG Foundation Models: A Systematic Study under Real-World Distribution Shiftshttps://blog.schoergendorfer.com/post/lightweight-genai-for-network-traffic-generation-fidelity-augmentation-and-classSecFeedde2026-08-26T07:47:26.133382+00:00Lightweight GenAI for Network Traffic Generation: Fidelity, Augmentation, and Classificationhttps://blog.schoergendorfer.com/post/beyond-oauth-task-scoped-authorization-for-ai-agents-via-natural-language-slicesSecFeedde2026-08-26T07:47:26.103822+00:00Beyond OAuth: Task-Scoped Authorization for AI Agents via Natural Language Sliceshttps://blog.schoergendorfer.com/post/msdata-a-millisecond-resolution-network-dataset-for-advancing-time-series-foundaSecFeedde2026-08-26T07:32:27.614070+00:00msData: A Millisecond-Resolution Network Dataset for Advancing Time Series Foundation Modelshttps://blog.schoergendorfer.com/post/adversa-measuring-multi-turn-guardrail-degradation-and-judge-reliability-in-largSecFeedde2026-08-26T07:32:27.582264+00:00ADVERSA: Measuring Multi-Turn Guardrail Degradation and Judge Reliability in Large Language Modelshttps://blog.schoergendorfer.com/post/estllm-enhancing-estonian-capabilities-in-multilingual-llms-via-continued-pretraSecFeedde2026-08-26T07:32:27.553929+00:00EstLLM: Enhancing Estonian Capabilities in Multilingual LLMs via Continued Pretraining and Post-Traininghttps://blog.schoergendorfer.com/post/st-lite-training-free-kv-cache-compression-with-spatio-trajectory-guidance-for-lSecFeedde2026-08-26T07:32:27.523522+00:00ST-Lite: Training-Free KV Cache Compression with Spatio-Trajectory Guidance for Long-Horizon GUI Agentshttps://blog.schoergendorfer.com/post/vlanext-recipes-for-building-strong-vla-modelsSecFeedde2026-08-26T07:32:27.493118+00:00VLANeXt: Recipes for Building Strong VLA Modelshttps://blog.schoergendorfer.com/post/you-can-learn-tokenization-end-to-end-with-reinforcement-learningSecFeedde2026-08-26T07:32:27.465641+00:00You Can Learn Tokenization End-to-End with Reinforcement Learninghttps://blog.schoergendorfer.com/post/patienthub-a-unified-framework-for-patient-simulationSecFeedde2026-08-26T07:32:27.433563+00:00PatientHub: A Unified Framework for Patient Simulationhttps://blog.schoergendorfer.com/post/ica-information-aware-credit-assignment-for-visually-grounded-long-horizon-inforSecFeedde2026-08-26T07:32:27.403397+00:00ICA: Information-Aware Credit Assignment for Visually Grounded Long-Horizon Information-Seeking Agentshttps://blog.schoergendorfer.com/post/anytime-pretraining-horizon-free-learning-rate-schedules-with-weight-averagingSecFeedde2026-08-26T07:32:27.372280+00:00Anytime Pretraining: Horizon-Free Learning-Rate Schedules with Weight Averaginghttps://blog.schoergendorfer.com/post/ad-insertion-in-llm-generated-responsesSecFeedde2026-08-26T07:32:27.342537+00:00Ad Insertion in LLM-Generated Responseshttps://blog.schoergendorfer.com/post/scientific-image-synthesis-benchmarking-methodologies-and-downstream-utilitySecFeedde2026-08-26T07:32:27.312533+00:00Scientific Image Synthesis: Benchmarking, Methodologies, and Downstream Utilityhttps://blog.schoergendorfer.com/post/tangrampuzzle-evaluating-multimodal-large-language-models-with-compositional-spaSecFeedde2026-08-26T07:32:27.280683+00:00TangramPuzzle: Evaluating Multimodal Large Language Models with Compositional Spatial Reasoninghttps://blog.schoergendorfer.com/post/minimal-decision-dynamics-and-contextual-probability-a-quantum-tug-of-war-modelSecFeedde2026-08-26T07:32:27.252433+00:00Minimal Decision Dynamics and Contextual Probability: A Quantum Tug-of-War Modelhttps://blog.schoergendorfer.com/post/seeing-vs-believing-evaluating-the-language-bias-of-open-source-mllms-in-counterSecFeedde2026-08-26T07:32:27.224674+00:00Seeing vs. Believing: Evaluating the Language Bias of Open-Source MLLMs in Counter-Intuitive Sceneshttps://blog.schoergendorfer.com/post/can-large-language-models-still-explain-themselves-investigating-the-impact-of-qSecFeedde2026-08-26T07:32:27.188337+00:00Can Large Language Models Still Explain Themselves? Investigating the Impact of Quantization on Self-Explanationshttps://blog.schoergendorfer.com/post/robust-motion-generation-using-part-level-reliable-data-from-videosSecFeedde2026-08-26T07:32:27.138330+00:00Robust Motion Generation using Part-level Reliable Data from Videoshttps://blog.schoergendorfer.com/post/cytonet-a-foundation-model-for-the-human-cerebral-cortex-at-cellular-resolutionSecFeedde2026-08-26T07:32:27.109457+00:00CytoNet: A Foundation Model for the Human Cerebral Cortex at Cellular Resolutionhttps://blog.schoergendorfer.com/post/monotone-and-separable-set-functions-characterizations-and-neural-models-2SecFeedde2026-08-26T07:32:27.083061+00:00Monotone and Separable Set Functions: Characterizations and Neural Modelshttps://blog.schoergendorfer.com/post/do-joint-language-audio-embeddings-encode-perceptual-timbre-semanticsSecFeedde2026-08-26T07:17:49.892625+00:00Do Joint Language-Audio Embeddings Encode Perceptual Timbre Semantics?https://blog.schoergendorfer.com/post/vggt-dp-generalizable-robot-control-via-vision-foundation-modelsSecFeedde2026-08-26T07:17:49.864309+00:00VGGT-DP: Generalizable Robot Control via Vision Foundation Modelshttps://blog.schoergendorfer.com/post/review-of-explainable-decision-support-and-adaptive-human-machine-interfaces-forSecFeedde2026-08-26T07:17:49.833356+00:00Review of Explainable Decision Support and Adaptive Human-Machine Interfaces for Automation Transparency in Maritime Autonomous Surface Shipshttps://blog.schoergendorfer.com/post/sta-net-a-decoupled-shape-and-texture-attention-network-for-lightweight-plant-diSecFeedde2026-08-26T07:17:49.808494+00:00STA-Net: A Decoupled Shape and Texture Attention Network for Lightweight Plant Disease Classificationhttps://blog.schoergendorfer.com/post/an-information-flow-perspective-on-explainability-requirements-specification-and-2SecFeedde2026-08-26T07:17:49.787322+00:00An Information-Flow Perspective on Explainability Requirements: Specification and Verificationhttps://blog.schoergendorfer.com/post/neurontune-fine-grained-neuron-modulation-for-balanced-safety-utility-alignmentSecFeedde2026-08-26T07:17:49.759583+00:00NeuronTune: Fine-Grained Neuron Modulation for Balanced Safety-Utility Alignment in LLMshttps://blog.schoergendorfer.com/post/can-large-language-models-assist-choice-modelling-insights-into-prompting-strateSecFeedde2026-08-26T07:17:49.730325+00:00Can large language models assist choice modelling? Insights into prompting strategies and current models' capabilitieshttps://blog.schoergendorfer.com/post/from-empirical-evaluation-to-context-aware-enhancement-repairing-regression-erroSecFeedde2026-08-26T07:17:49.701487+00:00From Empirical Evaluation to Context-Aware Enhancement: Repairing Regression Errors with LLMshttps://blog.schoergendorfer.com/post/a-modular-multitask-reasoning-framework-integrating-spatio-temporal-models-and-l-2SecFeedde2026-08-26T07:17:49.675818+00:00A Modular Multitask Reasoning Framework Integrating Spatio-temporal Models and LLMshttps://blog.schoergendorfer.com/post/quasar-a-programming-language-specialized-for-llm-code-actionsSecFeedde2026-08-26T07:17:49.648702+00:00Quasar: A Programming Language Specialized for LLM Code Actionshttps://blog.schoergendorfer.com/post/comparing-uncertainty-measurement-and-mitigation-methods-for-large-language-modeSecFeedde2026-08-26T07:17:49.594107+00:00Comparing Uncertainty Measurement and Mitigation Methods for Large Language Models: A Systematic Reviewhttps://blog.schoergendorfer.com/post/generative-ai-for-validating-physics-lawsSecFeedde2026-08-26T07:17:49.563472+00:00Generative AI for Validating Physics Lawshttps://blog.schoergendorfer.com/post/highway-congestion-reduction-through-reinforcement-learning-based-eulerian-headwSecFeedde2026-08-26T07:17:49.534606+00:00Highway Congestion Reduction through Reinforcement Learning Based Eulerian Headway Controlhttps://blog.schoergendorfer.com/post/intrinsic-papr-tackling-misattribution-in-3d-intrinsic-decomposition-via-proximiSecFeedde2026-08-26T07:17:49.507224+00:00Intrinsic PAPR: Tackling Misattribution in 3D Intrinsic Decomposition via Proximity Attention Point Renderinghttps://blog.schoergendorfer.com/post/screening-autism-spectrum-disorder-in-children-using-deep-learning-approach-evalSecFeedde2026-08-26T07:17:49.444662+00:00Screening Autism Spectrum Disorder in children using Deep Learning Approach : Evaluating the classification model of YOLOv26s by comparing with other modelshttps://blog.schoergendorfer.com/post/mediskill-evo-process-constrained-self-evolution-for-evidence-grounded-clinical-2SecFeedde2026-08-26T07:17:49.416485+00:00MediSkill-Evo: Process-Constrained Self-Evolution for Evidence-Grounded Clinical Interactionhttps://blog.schoergendorfer.com/post/apodex-11-scaling-agentic-intelligence-for-complex-work-2SecFeedde2026-08-26T07:17:49.382478+00:00Apodex 1.1: Scaling Agentic Intelligence for Complex Workhttps://blog.schoergendorfer.com/post/mobilepa-bench-benchmarking-mobile-planner-agents-on-complex-real-world-tasks-2SecFeedde2026-08-26T07:17:49.355535+00:00MobilePA-Bench: Benchmarking Mobile Planner Agents on Complex Real-World Taskshttps://blog.schoergendorfer.com/post/causalcache-conditional-high-fidelity-restoration-for-long-horizon-gui-agents-2SecFeedde2026-08-26T07:03:09.663644+00:00CausalCache: Conditional High-Fidelity Restoration for Long-Horizon GUI Agentshttps://blog.schoergendorfer.com/post/beyond-what-meets-the-eye-unveiling-situational-illusions-for-multimodal-large-l-2SecFeedde2026-08-26T07:03:09.632597+00:00Beyond What Meets the Eye: Unveiling Situational Illusions for Multimodal Large Language Modelshttps://blog.schoergendorfer.com/post/gencoord-skill-path-commitments-under-private-information-2SecFeedde2026-08-26T07:03:09.603128+00:00GenCoord: Skill-Path Commitments under Private Informationhttps://blog.schoergendorfer.com/post/what-you-cant-see-is-what-you-learn-slot-selective-evidence-masking-favors-compoSecFeedde2026-08-26T07:03:09.537182+00:00What You Can't See Is What You Learn: Slot-Selective Evidence Masking Favors Compositional Generalization in Shared-Genome Language-Model Societieshttps://blog.schoergendorfer.com/post/exphy-a-benchmark-for-explicit-physical-property-learning-in-multi-object-trajec-2SecFeedde2026-08-26T07:03:09.503161+00:00ExPhy: A Benchmark for Explicit Physical Property Learning in Multi-Object Trajectory Forecastinghttps://blog.schoergendorfer.com/post/reconstruction-a-blind-benchmark-for-recovering-research-ideas-from-pre-publicat-3SecFeedde2026-08-26T07:03:09.475235+00:00Reconstruction: A Blind Benchmark for Recovering Research Ideas from Pre-Publication Bibliographieshttps://blog.schoergendorfer.com/post/model-discovery-agent-llm-assisted-bayesian-experiment-design-for-data-efficient-4SecFeedde2026-08-26T07:03:09.402129+00:00Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world modelshttps://blog.schoergendorfer.com/post/superlocalmemory-40-the-governed-memory-operating-system-for-ai-agents-2SecFeedde2026-08-26T07:03:09.375558+00:00SuperLocalMemory 4.0: The Governed Memory Operating System for AI Agentshttps://blog.schoergendorfer.com/post/atomic-units-of-x-the-compression-layer-of-intelligence-2SecFeedde2026-08-26T07:03:09.351483+00:00Atomic Units of X: The Compression Layer of Intelligencehttps://blog.schoergendorfer.com/post/self-evolving-scientific-agent-designs-physically-reasoned-whitebox-fluid-controSecFeedde2026-08-26T07:03:09.325910+00:00Self-Evolving Scientific Agent Designs Physically-Reasoned Whitebox Fluid Controlhttps://blog.schoergendorfer.com/post/engiai-capability-based-evaluation-of-tool-connected-llm-agents-for-engineeringSecFeedde2026-08-26T07:03:09.298412+00:00EngiAI: Capability-Based Evaluation of Tool-Connected LLM Agents for Engineering Designhttps://blog.schoergendorfer.com/post/strategic-exploitation-in-llm-agent-markets-a-simulation-framework-for-e-commercSecFeedde2026-08-26T07:03:09.271749+00:00Strategic Exploitation in LLM Agent Markets: A Simulation Framework for E-Commerce Trusthttps://blog.schoergendorfer.com/post/reactbench-a-benchmark-for-topological-reasoning-in-mllms-on-chemical-reaction-dSecFeedde2026-08-26T07:03:09.227899+00:00ReactBench: A Benchmark for Topological Reasoning in MLLMs on Chemical Reaction Diagramshttps://blog.schoergendorfer.com/post/phmforge-evaluating-llm-agents-on-industrial-prognostics-through-mcp-native-algoSecFeedde2026-08-26T07:03:09.179738+00:00PHMForge: Evaluating LLM Agents on Industrial Prognostics through MCP-Native, Algorithm-Grounded Toolshttps://blog.schoergendorfer.com/post/comma-contribution-aware-medical-multi-agents-for-decentralized-oncology-decisioSecFeedde2026-08-26T06:47:53.145930+00:00CoMMa: Contribution-Aware Medical Multi-Agents for Decentralized Oncology Decision Supporthttps://blog.schoergendorfer.com/post/comparing-explanations-is-not-enough-explain-the-change-new-standards-are-neededSecFeedde2026-08-26T06:47:53.120625+00:00Comparing Explanations is Not Enough, Explain the Change: New Standards are Needed to Explain Behavioral Shifts in Large Language Modelshttps://blog.schoergendorfer.com/post/panning-for-gold-expanding-domain-specific-knowledge-graphs-with-general-knowledSecFeedde2026-08-26T06:47:53.095246+00:00Panning for Gold: Expanding Domain-Specific Knowledge Graphs with General Knowledgehttps://blog.schoergendorfer.com/post/reflctrl-controlling-llm-reflection-efficiently-via-representation-engineeringSecFeedde2026-08-26T06:47:53.068310+00:00ReflCtrl: Controlling LLM Reflection Efficiently via Representation Engineeringhttps://blog.schoergendorfer.com/post/uco-a-multi-turn-interactive-reinforcement-learning-method-for-adaptive-teachingSecFeedde2026-08-26T06:47:53.041991+00:00UCO: A Multi-Turn Interactive Reinforcement Learning Method for Adaptive Teaching with Large Language Modelshttps://blog.schoergendorfer.com/post/adaptive-gr1-specification-repair-for-liveness-preserving-shielding-in-reinforceSecFeedde2026-08-26T06:47:53.013287+00:00Adaptive GR(1) Specification Repair for Liveness-Preserving Shielding in Reinforcement Learninghttps://blog.schoergendorfer.com/post/illuminating-the-three-dogmas-of-reinforcement-learning-under-evolutionary-lightSecFeedde2026-08-26T06:47:52.986363+00:00Illuminating the Three Dogmas of Reinforcement Learning under Evolutionary Lighthttps://blog.schoergendorfer.com/post/efficient-llm-collaboration-via-planningSecFeedde2026-08-26T06:47:52.960839+00:00Efficient LLM Collaboration via Planninghttps://blog.schoergendorfer.com/post/olapa-mcot-enhancing-the-chinese-mathematical-reasoning-capability-of-llmsSecFeedde2026-08-26T06:47:52.910288+00:00Olapa-MCoT: Enhancing the Chinese Mathematical Reasoning Capability of LLMshttps://blog.schoergendorfer.com/post/topology-guided-modular-actor-critic-learning-for-continuous-systems-under-tempoSecFeedde2026-08-26T06:47:52.887043+00:00Topology-Guided Modular Actor-Critic Learning for Continuous Systems under Temporal Objectiveshttps://blog.schoergendorfer.com/post/fuzzy-segmentations-of-a-stringSecFeedde2026-08-26T06:47:52.862444+00:00Fuzzy Segmentations of a Stringhttps://blog.schoergendorfer.com/post/laion-bvd-a-10-million-hour-open-video-dataset-for-multimodal-pre-trainingSecFeedde2026-08-26T06:47:52.839190+00:00LAION-BVD: A 10-Million-Hour Open Video Dataset for Multimodal Pre-traininghttps://blog.schoergendorfer.com/post/reading-is-not-using-retrieval-judgment-and-the-design-of-ai-financial-researchSecFeedde2026-08-26T06:47:52.814466+00:00Reading Is Not Using: Retrieval, Judgment, and the Design of AI Financial Research Workflowshttps://blog.schoergendorfer.com/post/automatic-model-card-generation-using-an-llmSecFeedde2026-08-26T06:47:52.788332+00:00Automatic Model Card Generation Using an LLMhttps://blog.schoergendorfer.com/post/ensemble-of-convolutional-neural-networks-for-strokeprediction-towards-improvedSecFeedde2026-08-26T06:47:52.761712+00:00Ensemble of Convolutional Neural Networks for StrokePrediction: Towards Improved Diagnostic Accuracyhttps://blog.schoergendorfer.com/post/score-based-ideal-observer-approximation-via-denoising-score-matching-for-signalSecFeedde2026-08-26T06:47:52.736768+00:00Score-Based Ideal Observer Approximation via Denoising Score Matching for Signal-Known-Exactly Detection Taskshttps://blog.schoergendorfer.com/post/the-rat-a-unified-bayesian-model-for-rag-evaluationSecFeedde2026-08-26T06:47:52.678762+00:00The RAT: A Unified Bayesian Model for RAG Evaluationhttps://blog.schoergendorfer.com/post/method-mind-and-morality-how-people-make-sense-of-artificial-intelligenceSecFeedde2026-08-26T06:47:52.649626+00:00Method, Mind, and Morality: How People Make Sense of Artificial Intelligencehttps://blog.schoergendorfer.com/post/parameter-efficient-self-supervised-adaptation-for-eeg-fm-under-fixed-computatioSecFeedde2026-08-26T06:32:34.527307+00:00Parameter-Efficient Self-Supervised Adaptation for EEG-FM under Fixed Computational Budgetshttps://blog.schoergendorfer.com/post/enhancing-bayesian-optimization-and-active-learning-through-kernel-diversitySecFeedde2026-08-26T06:32:34.499077+00:00Enhancing Bayesian Optimization and Active Learning Through Kernel Diversityhttps://blog.schoergendorfer.com/post/deep-learning-super-resolution-for-satellite-cloud-mask-downscalingSecFeedde2026-08-26T06:32:34.476057+00:00Deep Learning Super Resolution for Satellite Cloud Mask Downscalinghttps://blog.schoergendorfer.com/post/constrained-hyperparameter-optimization-for-streaming-dataSecFeedde2026-08-26T06:32:34.450476+00:00Constrained Hyperparameter Optimization for Streaming Datahttps://blog.schoergendorfer.com/post/on-policy-distillation-with-verifiable-rewardSecFeedde2026-08-26T06:32:34.427487+00:00On-policy Distillation with Verifiable Rewardhttps://blog.schoergendorfer.com/post/simthesizer-an-agent-driven-simulation-framework-for-llm-serving-systemsSecFeedde2026-08-26T06:32:34.381808+00:00Simthesizer: An Agent-Driven Simulation Framework for LLM Serving Systemshttps://blog.schoergendorfer.com/post/taming-foundation-model-with-invariance-oriented-pre-training-for-broad-spectrumSecFeedde2026-08-26T06:32:34.327483+00:00Taming foundation model with invariance-oriented pre-training for broad-spectrum EEG analysis across signal-level, brain-state, and brain-health taskshttps://blog.schoergendorfer.com/post/textttfindr-transparent-and-fair-credit-risk-decisions-through-semi-structured-rSecFeedde2026-08-26T06:32:34.302719+00:00$\texttt{findr}$: Transparent and Fair Credit Risk Decisions through Semi-Structured Regressionshttps://blog.schoergendorfer.com/post/across-the-loss-landscape-with-progressive-growthSecFeedde2026-08-26T06:32:34.268541+00:00Across the Loss Landscape with Progressive Growthhttps://blog.schoergendorfer.com/post/strokeguard-a-multi-agent-guided-system-for-prehospital-stroke-assessmentSecFeedde2026-08-26T06:32:34.209730+00:00StrokeGuard: A Multi-Agent Guided System for Prehospital Stroke Assessmenthttps://blog.schoergendorfer.com/post/fraudbench-protocol-sensitive-benchmarking-of-adversarial-robustness-for-financiSecFeedde2026-08-26T06:32:34.178062+00:00FraudBench: Protocol-Sensitive Benchmarking of Adversarial Robustness for Financial Risk Assessmenthttps://blog.schoergendorfer.com/post/lumixai-a-modular-full-stack-framework-for-feature-attributionSecFeedde2026-08-26T06:32:34.150134+00:00LumiXAI: A Modular Full-Stack Framework for Feature Attributionhttps://blog.schoergendorfer.com/post/scalable-and-versatile-identification-for-hierarchical-structural-causal-modelsSecFeedde2026-08-26T06:32:34.123515+00:00Scalable and Versatile Identification for Hierarchical Structural Causal Models: A New Look at Project STARhttps://blog.schoergendorfer.com/post/when-do-supervised-uq-ensembles-improve-llm-hallucination-detection-a-robustnessSecFeedde2026-08-26T06:32:34.098180+00:00When Do Supervised UQ Ensembles Improve LLM Hallucination Detection? A Robustness Studyhttps://blog.schoergendorfer.com/post/beyond-static-interpretability-anticipating-post-sft-mechanisms-from-pre-sft-parSecFeedde2026-08-26T06:32:34.073675+00:00Beyond Static Interpretability: Anticipating Post-SFT Mechanisms from Pre-SFT Parameters for Better Tuninghttps://blog.schoergendorfer.com/post/evaluating-deep-multivariate-imputation-models-on-wearable-device-dataSecFeedde2026-08-26T06:32:34.047923+00:00Evaluating Deep Multivariate Imputation Models on Wearable Device Datahttps://blog.schoergendorfer.com/post/multilevel-fair-allocation-under-additive-preferencesSecFeedde2026-08-26T06:32:34.018063+00:00Multilevel Fair Allocation under Additive Preferenceshttps://blog.schoergendorfer.com/post/equivariant-covariance-tensors-guaranteed-spd-uncertainty-for-tensor-valued-geomSecFeedde2026-08-26T06:17:29.709725+00:00Equivariant Covariance Tensors: Guaranteed SPD Uncertainty for Tensor-Valued Geometric Learninghttps://blog.schoergendorfer.com/post/markerless-pose-estimation-for-resistance-training-technique-assessmentSecFeedde2026-08-26T06:17:29.675923+00:00Markerless Pose Estimation for Resistance Training Technique Assessmenthttps://blog.schoergendorfer.com/post/not-all-tokens-are-equal-region-aware-consistency-repair-of-backdoors-in-mllmsSecFeedde2026-08-26T06:17:29.643801+00:00Not All Tokens Are Equal: Region-Aware Consistency Repair of Backdoors in MLLMshttps://blog.schoergendorfer.com/post/farca-fact-aligned-reliability-aware-credit-assignment-for-reinforcement-learninSecFeedde2026-08-26T06:17:29.616177+00:00FARCA: Fact-Aligned Reliability-Aware Credit Assignment for Reinforcement Learning with Factual Supervisionhttps://blog.schoergendorfer.com/post/metadata-aware-adaptation-of-a-generative-foundation-model-for-conditional-cmr-sSecFeedde2026-08-26T06:17:29.587131+00:00Metadata-Aware Adaptation of a Generative Foundation Model for Conditional CMR Synthesishttps://blog.schoergendorfer.com/post/mind-the-student-behavioral-and-contextual-cues-for-automated-engagement-predictSecFeedde2026-08-26T06:17:29.554986+00:00Mind the Student: Behavioral and Contextual Cues for Automated Engagement Prediction in Online Learninghttps://blog.schoergendorfer.com/post/senseshift-continuous-sentiment-controlled-text-generation-via-encoder-based-masSecFeedde2026-08-26T06:17:29.527721+00:00SENSESHIFT: Continuous Sentiment-Controlled Text Generation via Encoder-based Mask Infillinghttps://blog.schoergendorfer.com/post/contrastive-branch-policy-optimizationSecFeedde2026-08-26T06:17:29.502729+00:00Contrastive Branch Policy Optimizationhttps://blog.schoergendorfer.com/post/ghaib-in-translation-aka-unseen-harm-measuring-cross-script-safety-inconsistencySecFeedde2026-08-26T06:17:29.474905+00:00'Ghaib in Translation' aka Unseen Harm: Measuring Cross-Script Safety Inconsistency with 'Missed-in-Urdu' Scores in LLM Hate Speech Detectionhttps://blog.schoergendorfer.com/post/tlow-flow-based-item-tokenizer-for-recommendationSecFeedde2026-08-26T06:17:29.445736+00:00Tlow: Flow-based Item Tokenizer for Recommendationhttps://blog.schoergendorfer.com/post/llm-guided-contextual-action-evaluation-for-operational-decisions-in-industrialSecFeedde2026-08-26T06:17:29.386507+00:00LLM-Guided Contextual Action Evaluation for Operational Decisions in Industrial Processeshttps://blog.schoergendorfer.com/post/rethinking-pre-training-and-augmentation-for-zero-shot-cross-city-object-detectiSecFeedde2026-08-26T06:17:29.359566+00:00Rethinking Pre-Training and Augmentation for Zero-Shot Cross-City Object Detectionhttps://blog.schoergendorfer.com/post/placeseek-human-centered-geospatial-retrieval-of-urban-outdoor-places-via-semantSecFeedde2026-08-26T06:17:29.326443+00:00PlaceSeek: Human-Centered Geospatial Retrieval of Urban Outdoor Places via Semantic Grounding and Affective Alignmenthttps://blog.schoergendorfer.com/post/from-gradient-boosted-trees-to-deep-recommenders-practical-lessons-from-migratinSecFeedde2026-08-26T06:17:29.300263+00:00From Gradient-Boosted Trees to Deep Recommenders: Practical Lessons from Migrating a Production Customer Support Recommenderhttps://blog.schoergendorfer.com/post/syn2realtrack-bridging-the-gap-between-synthetic-and-real-world-datasets-for-onlSecFeedde2026-08-26T06:17:29.267829+00:00Syn2RealTrack: Bridging the Gap Between Synthetic and Real-World Datasets for Online Multi-View Multi-Target Trackinghttps://blog.schoergendorfer.com/post/transphy-visual-in-context-learning-for-physically-grounded-image-editingSecFeedde2026-08-26T06:17:29.242046+00:00TransPhy: Visual In-Context Learning for Physically Grounded Image Editinghttps://blog.schoergendorfer.com/post/ponderpounce-a-pretrained-mllm-as-an-episode-context-engine-for-robot-controlSecFeedde2026-08-26T06:17:29.215045+00:00PonderPounce: A Pretrained MLLM as an Episode Context Engine for Robot Controlhttps://blog.schoergendorfer.com/post/matreplace-a-reference-free-conditioning-aligned-benchmark-for-material-replacemSecFeedde2026-08-26T06:17:29.159185+00:00MatReplace: A Reference-Free, Conditioning-Aligned Benchmark for Material Replacement in Interior Sceneshttps://blog.schoergendorfer.com/post/knowing-when-to-ask-for-help-bayesian-self-escalation-in-hierarchical-llm-agentsSecFeedde2026-08-26T06:02:26.105548+00:00Knowing When to Ask for Help: Bayesian Self-Escalation in Hierarchical LLM Agentshttps://blog.schoergendorfer.com/post/partab-partition-aware-reasoning-with-structured-evidence-for-scalable-table-undSecFeedde2026-08-26T06:02:26.073157+00:00PARTAB: Partition-Aware Reasoning with Structured Evidence for Scalable Table Understandinghttps://blog.schoergendorfer.com/post/when-less-is-more-an-empirical-study-of-minimal-responses-in-counseling-dialogueSecFeedde2026-08-26T06:02:26.047051+00:00When Less Is More: An Empirical Study of Minimal Responses in Counseling Dialogues and the Behavior of LLMshttps://blog.schoergendorfer.com/post/orbitalif-an-efficient-spiking-federated-learning-framework-for-onboard-cloud-reSecFeedde2026-08-26T06:02:26.019954+00:00ORBITALIF: An Efficient Spiking Federated Learning Framework for Onboard Cloud Removalhttps://blog.schoergendorfer.com/post/mechanistic-circuit-identification-for-controllable-data-generationSecFeedde2026-08-26T06:02:25.984828+00:00Mechanistic Circuit Identification for Controllable Data Generationhttps://blog.schoergendorfer.com/post/viscache-visual-kv-cache-pruning-for-efficient-vision-large-language-model-inferSecFeedde2026-08-26T06:02:25.957860+00:00VisCache: Visual KV Cache Pruning for Efficient Vision Large Language Model Inferencehttps://blog.schoergendorfer.com/post/hierarchical-skill-retrieval-for-data-efficient-adaptation-of-vision-language-acSecFeedde2026-08-26T06:02:25.899224+00:00Hierarchical Skill Retrieval for Data-Efficient Adaptation of Vision-Language-Action Modelshttps://blog.schoergendorfer.com/post/design-to-plan-a-large-language-model-based-multi-agent-framework-for-manufacturSecFeedde2026-08-26T06:02:25.867952+00:00Design-to-Plan: A Large Language Model-Based Multi-Agent Framework for Manufacturing Process Planning from 3D CAD Models and 2D Engineering Drawingshttps://blog.schoergendorfer.com/post/chorustic-training-free-multivariate-time-series-classification-via-chorus-in-coSecFeedde2026-08-26T06:02:25.835645+00:00ChorusTIC: Training-Free Multivariate Time Series Classification via Chorus In-Context Learninghttps://blog.schoergendorfer.com/post/what-guides-the-agent-adjudicating-unauthorized-behavior-via-localizing-behaviorSecFeedde2026-08-26T06:02:25.806864+00:00What Guides the Agent? Adjudicating Unauthorized Behavior via Localizing Behavior-Guiding Instructionshttps://blog.schoergendorfer.com/post/itercad-iterative-program-repair-for-cad-code-generation-from-orthographic-viewsSecFeedde2026-08-26T06:02:25.778892+00:00IterCAD: Iterative Program Repair for CAD Code Generation from Orthographic Viewshttps://blog.schoergendorfer.com/post/webmcp-phalanx-enforcing-and-characterizing-trust-boundaries-for-browser-integraSecFeedde2026-08-26T06:02:25.749370+00:00WebMCP-Phalanx: Enforcing and Characterizing Trust Boundaries for Browser-Integrated LLM Agentshttps://blog.schoergendorfer.com/post/hybrid-semantic-tool-discovery-for-enterprise-mcp-gateway-architecture-and-impleSecFeedde2026-08-26T06:02:25.698804+00:00Hybrid Semantic Tool Discovery for Enterprise MCP Gateway: Architecture and Implementationhttps://blog.schoergendorfer.com/post/the-shadow-price-of-intelligence-quality-degradation-in-llm-inference-as-a-supplSecFeedde2026-08-26T06:02:25.672411+00:00The Shadow Price of Intelligence: Quality Degradation in LLM Inference as a Supply Chain Problemhttps://blog.schoergendorfer.com/post/ragsentinel-certifiable-geometric-consensus-for-robust-retrieval-augmented-generSecFeedde2026-08-26T06:02:25.642933+00:00RAGSentinel: Certifiable Geometric Consensus for Robust Retrieval-Augmented Generationhttps://blog.schoergendorfer.com/post/evaluating-language-models-on-cross-language-code-functional-equivalenceSecFeedde2026-08-26T06:02:25.616878+00:00Evaluating Language Models on Cross-Language Code Functional Equivalencehttps://blog.schoergendorfer.com/post/neuronguard-robust-llm-safety-alignment-via-ablation-aware-safety-signal-redistrSecFeedde2026-08-26T06:02:25.585583+00:00NeuronGuard: Robust LLM Safety Alignment via Ablation-Aware Safety Signal Redistributionhttps://blog.schoergendorfer.com/post/the-empire-long-divided-must-unite-architectural-convergence-in-three-llm-agentSecFeedde2026-08-26T06:02:25.542535+00:00The Empire, Long Divided, Must Unite: Architectural Convergence in Three LLM Agent Harnesseshttps://blog.schoergendorfer.com/post/stain-fl-stealthy-targeted-attack-injection-with-contextual-triggers-in-federateSecFeedde2026-08-26T05:47:37.743458+00:00STAIN-FL: Stealthy Targeted Attack Injection with Contextual Triggers in Federated Learninghttps://blog.schoergendorfer.com/post/luce-relightable-gaussians-for-3d-asset-generationSecFeedde2026-08-26T05:47:37.715701+00:00Luce: Relightable Gaussians for 3D Asset Generationhttps://blog.schoergendorfer.com/post/qml-for-quantum-sensing-under-measurement-induced-information-lossSecFeedde2026-08-26T05:47:37.689145+00:00QML for Quantum Sensing under Measurement-Induced Information Losshttps://blog.schoergendorfer.com/post/refinerank-joint-box-refinement-and-ranking-for-surgical-spatio-temporal-groundiSecFeedde2026-08-26T05:47:37.660155+00:00RefineRank: Joint Box Refinement and Ranking for Surgical Spatio-Temporal Groundinghttps://blog.schoergendorfer.com/post/names-can-hurt-spotting-slopsquatting-risks-caused-by-package-name-hallucinationSecFeedde2026-08-26T05:47:37.636738+00:00Names Can Hurt: Spotting Slopsquatting Risks Caused by Package Name Hallucinations in Local Coding LLMshttps://blog.schoergendorfer.com/post/learning-the-kohn-sham-map-with-neural-operators-for-quasi-linear-scaling-densitSecFeedde2026-08-26T05:47:37.613105+00:00Learning the Kohn-Sham map with neural operators for quasi-linear scaling density functional theoryhttps://blog.schoergendorfer.com/post/a-mathematical-theory-of-interpretation-rational-entropy-spectral-readout-and-coSecFeedde2026-08-26T05:47:37.584565+00:00A Mathematical Theory of Interpretation: Rational Entropy, Spectral Readout, and Confusability as a Resourcehttps://blog.schoergendorfer.com/post/a-tale-of-perfect-fit-and-phantom-optima-how-data-driven-models-can-fail-in-realSecFeedde2026-08-26T05:47:37.559682+00:00A tale of perfect fit and phantom optima: how data-driven models can fail in real-time optimizationhttps://blog.schoergendorfer.com/post/revelation-controlSecFeedde2026-08-26T05:47:37.532623+00:00Revelation Controlhttps://blog.schoergendorfer.com/post/beyond-the-mandate-a-systematic-security-analysis-of-the-agent-payments-protocolSecFeedde2026-08-26T05:47:37.505896+00:00Beyond the Mandate: A Systematic Security Analysis of the Agent Payments Protocol (AP2)https://blog.schoergendorfer.com/post/automated-synthesis-of-cloud-emulatorsSecFeedde2026-08-26T05:47:37.452057+00:00Automated Synthesis of Cloud Emulatorshttps://blog.schoergendorfer.com/post/resilience-matters-for-embodied-agents-system-new-metrics-systematic-evaluationSecFeedde2026-08-26T05:47:37.393185+00:00Resilience Matters for Embodied Agents System: New Metrics, Systematic Evaluation, and Optimizationhttps://blog.schoergendorfer.com/post/infant-care-video-dataset-for-classification-of-interventions-using-transformersSecFeedde2026-08-26T05:47:37.363150+00:00Infant Care Video Dataset for Classification of Interventions Using Transformershttps://blog.schoergendorfer.com/post/place-slice-and-schedule-hierarchical-o-ran-control-of-a-tethered-mmwave-uav-gnbSecFeedde2026-08-26T05:47:37.310540+00:00Place, Slice and Schedule: Hierarchical O-RAN Control of a Tethered mmWave UAV-gNBhttps://blog.schoergendorfer.com/post/learning-to-grade-efficiently-a-bandit-driven-prompt-selection-framework-for-lowSecFeedde2026-08-26T05:47:37.283499+00:00Learning to Grade Efficiently: A Bandit-Driven Prompt-Selection Framework for Low-Cost LLM Essay Scoringhttps://blog.schoergendorfer.com/post/discovering-cross-language-reasoning-invariance-in-llms-with-geometry-invariantSecFeedde2026-08-26T05:47:37.255057+00:00Discovering Cross-Language Reasoning Invariance in LLMs with Geometry-Invariant Sparse Autoencodershttps://blog.schoergendorfer.com/post/lucaid-agentic-multimodal-ai-for-lung-cancer-precision-pathologySecFeedde2026-08-26T05:47:37.228577+00:00LUCAID: Agentic Multimodal AI for Lung Cancer Precision Pathologyhttps://blog.schoergendorfer.com/post/restoring-without-forgetting-continual-learning-across-image-degradationsSecFeedde2026-08-26T05:32:28.565331+00:00Restoring Without Forgetting: Continual Learning Across Image Degradationshttps://blog.schoergendorfer.com/post/emotra-tts-smooth-intra-utterance-emotion-transitions-for-speech-synthesisSecFeedde2026-08-26T05:32:28.537352+00:00EmoTra-TTS: Smooth Intra-Utterance Emotion Transitions for Speech Synthesishttps://blog.schoergendorfer.com/post/when-youth-enter-the-chat-an-epistemic-shift-in-the-validation-of-llm-based-measSecFeedde2026-08-26T05:32:28.517800+00:00When Youth Enter The Chat: An Epistemic Shift in the Validation of LLM-Based Measures of Student Talkhttps://blog.schoergendorfer.com/post/disentangled-skill-representations-for-predictive-human-modelingSecFeedde2026-08-26T05:32:28.496694+00:00Disentangled Skill Representations for Predictive Human Modelinghttps://blog.schoergendorfer.com/post/what-reaches-expert-review-representation-structural-screening-and-candidate-forSecFeedde2026-08-26T05:32:28.473723+00:00What Reaches Expert Review? Representation, Structural Screening, and Candidate-Form Dependence in AI-Assisted Item Developmenthttps://blog.schoergendorfer.com/post/trustshiftprobe-characterizing-benchmarking-and-defending-staged-trust-attacks-oSecFeedde2026-08-26T05:32:28.452025+00:00TrustShiftProbe: Characterizing, Benchmarking, and Defending Staged Trust Attacks on MCP Servershttps://blog.schoergendorfer.com/post/exam2-underlineextending-underlineaudio-understanding-in-underlinemultilingual-aSecFeedde2026-08-26T05:32:28.428985+00:00EXAM$^2$: $\underline{Ex}tending$ $\underline{A}udio$ $Understanding$ $in$ $\underline{M}ultilingual$ $and$ $\underline{M}ultimodal$ $Analysis$https://blog.schoergendorfer.com/post/too-much-of-a-good-thing-when-knowledge-distillation-promotes-overfitting-and-hoSecFeedde2026-08-26T05:32:28.403177+00:00Too much of a good thing -- when knowledge distillation promotes overfitting, and how to avoid ithttps://blog.schoergendorfer.com/post/the-limits-of-automatic-evaluation-of-creativity-in-large-language-modelsSecFeedde2026-08-26T05:32:28.379410+00:00The Limits of Automatic Evaluation of Creativity in Large Language Modelshttps://blog.schoergendorfer.com/post/confidently-wrong-silently-so-auditing-undetectable-failures-of-a-deployed-on-deSecFeedde2026-08-26T05:32:28.353470+00:00Confidently Wrong, Silently So: Auditing Undetectable Failures of a Deployed On-Device Language Modelhttps://blog.schoergendorfer.com/post/from-causal-plausibility-to-causal-reliability-evaluating-llms-as-calibrated-dirSecFeedde2026-08-26T05:32:28.323983+00:00From Causal Plausibility to Causal Reliability: Evaluating LLMs as Calibrated Direct Causal-Edge Classifiershttps://blog.schoergendorfer.com/post/elastic-kv-cache-for-llm-servinga-working-reclamation-mechanism-and-why-chunkedSecFeedde2026-08-26T05:32:28.290664+00:00Elastic KV Cache for LLM Serving:A Working Reclamation Mechanism, and Why Chunked Prefill Already Closes the Gaphttps://blog.schoergendorfer.com/post/beyond-executable-models-the-pufibara-agent-harness-and-the-modelica-agent-workfSecFeedde2026-08-26T05:32:28.265813+00:00Beyond Executable Models: The Pufibara Agent Harness and the Modelica Agent Workflow Benchmark for Physical System Modelinghttps://blog.schoergendorfer.com/post/feedback-that-backfires-why-small-language-model-agents-repeat-the-call-they-jusSecFeedde2026-08-26T05:32:28.239413+00:00Feedback That Backfires: Why Small Language Model Agents Repeat the Call They Just Watched Failhttps://blog.schoergendorfer.com/post/toolrobustbench-stage-wise-perturbation-evaluation-and-failure-diagnosis-for-tooSecFeedde2026-08-26T05:32:28.216063+00:00ToolRobustBench: Stage-Wise Perturbation Evaluation and Failure Diagnosis for Tool-Calling Agentshttps://blog.schoergendorfer.com/post/macro-operator-generation-and-predicate-selection-for-tamp-operator-learningSecFeedde2026-08-26T05:32:28.191015+00:00Macro-Operator Generation and Predicate Selection for TAMP Operator Learninghttps://blog.schoergendorfer.com/post/when-may-an-agent-stop-evidence-carrying-termination-for-tool-using-llmsSecFeedde2026-08-26T05:32:28.167256+00:00When May an Agent Stop? Evidence-Carrying Termination for Tool-Using LLMshttps://blog.schoergendorfer.com/post/identifying-latent-declarative-representations-of-code-for-assisting-repositorySecFeedde2026-08-26T05:32:28.141264+00:00Identifying Latent Declarative Representations of Code for Assisting Repository Migrationhttps://blog.schoergendorfer.com/post/rebuild-dossier-mechanically-enforced-specs-for-agentic-app-rebuilds-and-what-moSecFeedde2026-08-26T05:32:28.114056+00:00Rebuild Dossier: Mechanically-Enforced Specs for Agentic App Rebuilds, and What Model-Tier Failures Revealhttps://blog.schoergendorfer.com/post/refine-a-multi-agent-llm-approach-for-evidence-guided-code-refactoringSecFeedde2026-08-26T05:32:28.082173+00:00REFINE: A Multi-Agent LLM Approach for Evidence-Guided Code Refactoringhttps://blog.schoergendorfer.com/post/fidelity-preference-not-demographic-preference-a-pixel-level-attribute-sensitiviSecFeedde2026-08-26T05:17:44.693101+00:00Fidelity Preference, Not Demographic Preference: A Pixel-Level Attribute-Sensitivity Audit of Image Aesthetic/Preference Scorershttps://blog.schoergendorfer.com/post/a-human-factors-guided-cognitive-model-of-visuospatial-complexity-in-embodied-acSecFeedde2026-08-26T05:17:44.664876+00:00A Human-Factors Guided Cognitive Model of Visuospatial Complexity in Embodied Active Visionhttps://blog.schoergendorfer.com/post/progressively-learning-heterogeneous-skills-in-a-unified-latent-spaceSecFeedde2026-08-26T05:17:44.637127+00:00Progressively Learning Heterogeneous Skills in a Unified Latent Spacehttps://blog.schoergendorfer.com/post/recursive-experiential-working-memory-evolution-for-long-horizon-agent-harnessesSecFeedde2026-08-26T05:17:44.610086+00:00Recursive Experiential-Working Memory Evolution for Long-Horizon Agent Harnesseshttps://blog.schoergendorfer.com/post/fedv-kgqa-multi-hop-question-answering-over-vertically-partitioned-knowledge-graSecFeedde2026-08-26T05:17:44.560437+00:00FedV-KGQA: Multi-Hop Question Answering over Vertically Partitioned Knowledge Graphshttps://blog.schoergendorfer.com/post/a-dual-dimensional-llm-framework-for-automated-item-incidental-content-similaritSecFeedde2026-08-26T05:17:44.529419+00:00A Dual-Dimensional LLM Framework for Automated Item Incidental Content Similarity Analysis in Large-Scale Assessmentshttps://blog.schoergendorfer.com/post/constrained-entity-selection-under-partial-knowledge-for-llm-based-knowledge-graSecFeedde2026-08-26T05:17:44.500471+00:00Constrained Entity Selection under Partial Knowledge for LLM-Based Knowledge Graph QAhttps://blog.schoergendorfer.com/post/strictly-causal-streaming-video-anomaly-detection-with-a-theoretically-groundedSecFeedde2026-08-26T05:17:44.469125+00:00Strictly Causal Streaming Video Anomaly Detection with a Theoretically-Grounded State-Space Corehttps://blog.schoergendorfer.com/post/starharness-evolving-harnesses-with-stratified-search-for-enterprise-environmentSecFeedde2026-08-26T05:17:44.441583+00:00StarHarness: Evolving Harnesses with Stratified Search for Enterprise Environmentshttps://blog.schoergendorfer.com/post/cafe-self-improving-search-agents-need-co-evolving-feedbackSecFeedde2026-08-26T05:17:44.407718+00:00CAFE: Self-Improving Search Agents Need Co-Evolving Feedbackhttps://blog.schoergendorfer.com/post/right-diagnoses-decorative-reasoninga-perturbation-audit-of-medical-chain-of-thoSecFeedde2026-08-26T05:17:44.380931+00:00Right Diagnoses, Decorative Reasoning:A Perturbation Audit of Medical Chain-of-Thoughthttps://blog.schoergendorfer.com/post/stepguard-learning-step-level-guardrails-with-scalable-supervision-and-safety-utSecFeedde2026-08-26T05:17:44.351814+00:00StepGuard: Learning Step-Level Guardrails with Scalable Supervision and Safety-Utility Balancinghttps://blog.schoergendorfer.com/post/evidence-blindness-in-direct-corpus-interaction-persistent-navigation-with-atlasSecFeedde2026-08-26T05:17:44.322190+00:00Evidence Blindness in Direct Corpus Interaction: Persistent Navigation with AtlasNavhttps://blog.schoergendorfer.com/post/metan-recursive-self-improvement-through-emergent-depthSecFeedde2026-08-26T05:17:44.266540+00:00Meta$^n$: Recursive Self-Improvement through Emergent Depthhttps://blog.schoergendorfer.com/post/lifted-model-construction-under-approximate-commutativitySecFeedde2026-08-26T05:17:44.236268+00:00Lifted Model Construction under Approximate Commutativityhttps://blog.schoergendorfer.com/post/confident-at-the-moment-of-action-belief-miscalibration-in-llm-play-under-hiddenSecFeedde2026-08-26T05:17:44.211209+00:00Confident at the moment of action: belief miscalibration in LLM play under hidden informationhttps://blog.schoergendorfer.com/post/the-invisible-editorial-layer-formalizing-undisclosed-inference-time-steering-prSecFeedde2026-08-26T05:17:44.182423+00:00The Invisible Editorial Layer: Formalizing Undisclosed Inference-Time Steering, Probability Placement, and the Attribution Problem in Deployed Language Modelshttps://blog.schoergendorfer.com/post/parason-revealing-subtask-and-trial-parallelism-in-llm-reasoningSecFeedde2026-08-26T05:17:44.151903+00:00Parason: Revealing Subtask and Trial Parallelism in LLM Reasoninghttps://blog.schoergendorfer.com/post/causal-modelling-of-support-interventions-for-student-competency-assessmentSecFeedde2026-08-26T05:02:31.465920+00:00Causal Modelling of Support Interventions for Student Competency Assessmenthttps://blog.schoergendorfer.com/post/pivot-and-station-multi-agent-path-finding-solvability-complexity-and-algorithmsSecFeedde2026-08-26T05:02:31.441237+00:00Pivot-and-Station Multi-Agent Path Finding: Solvability, Complexity, and Algorithmshttps://blog.schoergendorfer.com/post/physmllms-spatial-priors-for-unified-referring-segmentation-and-grounded-reasoniSecFeedde2026-08-26T05:02:31.414631+00:00PhysMLLMs: Spatial Priors for Unified Referring Segmentation and Grounded Reasoning of Images and Videoshttps://blog.schoergendorfer.com/post/joint-optimization-of-tool-creation-and-use-for-large-language-model-agentsSecFeedde2026-08-26T05:02:31.385632+00:00Joint Optimization of Tool Creation and Use for Large Language Model Agentshttps://blog.schoergendorfer.com/post/evidx-evidence-aware-active-diagnosis-with-scaffolded-llm-agentsSecFeedde2026-08-26T05:02:31.360572+00:00EviDx: Evidence-Aware Active Diagnosis with Scaffolded LLM Agentshttps://blog.schoergendorfer.com/post/when-must-becomes-maybe-constraint-weakening-in-llm-agent-workflowsSecFeedde2026-08-26T05:02:31.335534+00:00When "Must" Becomes "Maybe": Constraint Weakening in LLM Agent Workflowshttps://blog.schoergendorfer.com/post/discovering-adaptive-transmission-programs-for-collective-innovationSecFeedde2026-08-26T05:02:31.303791+00:00Discovering Adaptive Transmission Programs for Collective Innovationhttps://blog.schoergendorfer.com/post/neurosymbolic-alignment-for-physiologically-safe-clinical-language-modelsSecFeedde2026-08-26T05:02:31.279508+00:00Neurosymbolic Alignment for Physiologically-Safe Clinical Language Modelshttps://blog.schoergendorfer.com/post/peakbench-benchmarking-resource-aware-tool-invocation-in-llm-agentsSecFeedde2026-08-26T05:02:31.253997+00:00PeakBench: Benchmarking Resource-Aware Tool Invocation in LLM Agentshttps://blog.schoergendorfer.com/post/implicit-q-learning-bootstrapped-ant-colony-optimization-for-maritime-moving-tarSecFeedde2026-08-26T05:02:31.231517+00:00Implicit Q-learning-bootstrapped ant colony optimization for maritime moving-target observation scheduling with agile satelliteshttps://blog.schoergendorfer.com/post/reinforcement-learning-guided-evolutionary-policy-optimization-for-preference-adSecFeedde2026-08-26T05:02:31.207139+00:00Reinforcement Learning-Guided Evolutionary Policy Optimization for Preference-Adjustable Heterogeneous Agile Earth Observation Satellite Schedulinghttps://blog.schoergendorfer.com/post/mahalanobis-based-multi-head-attention-for-complex-state-propagationSecFeedde2026-08-26T05:02:31.158146+00:00Mahalanobis-Based Multi-Head Attention for Complex State Propagationhttps://blog.schoergendorfer.com/post/a-behavior-guided-online-probabilistic-forecasting-method-for-electric-vehicle-cSecFeedde2026-08-26T05:02:31.126299+00:00A Behavior-Guided Online Probabilistic Forecasting Method for Electric vehicle Charging Loadshttps://blog.schoergendorfer.com/post/partial-identification-under-causal-orders-by-linear-programmingSecFeedde2026-08-26T05:02:31.090777+00:00Partial Identification under Causal Orders by Linear Programminghttps://blog.schoergendorfer.com/post/a-judge-should-know-what-changedconstruct-validity-for-llm-as-a-judge-evaluationSecFeedde2026-08-26T05:02:31.066325+00:00A Judge Should Know What Changed:Construct Validity for LLM-as-a-Judge Evaluationhttps://blog.schoergendorfer.com/post/resispec-enhancing-multi-candidate-speculative-sampling-via-residual-distributioSecFeedde2026-08-26T05:02:31.039403+00:00ResiSpec: Enhancing Multi-Candidate Speculative Sampling via Residual Distribution Shapinghttps://blog.schoergendorfer.com/post/do-recipes-have-personas-characterizing-and-generating-creator-style-in-attributSecFeedde2026-08-26T05:02:31.005582+00:00Do Recipes Have Personas? Characterizing and Generating Creator Style in Attributed Procedural Graphshttps://blog.schoergendorfer.com/post/from-state-to-action-ooda-tool-for-reliable-multi-turn-tool-useSecFeedde2026-08-26T05:02:30.976632+00:00From State to Action: OODA-Tool for Reliable Multi-Turn Tool Usehttps://blog.schoergendorfer.com/post/adaptive-influence-graphs-for-failure-attribution-in-multi-agent-systemsSecFeedde2026-08-26T05:02:30.947361+00:00Adaptive Influence Graphs for Failure Attribution in Multi-Agent Systemshttps://blog.schoergendorfer.com/post/the-handoff-tax-continuing-non-native-trajectories-in-llm-agentsSecFeedde2026-08-26T04:47:26.007586+00:00The Handoff Tax: Continuing Non-Native Trajectories in LLM Agentshttps://blog.schoergendorfer.com/post/selective-regenerative-decoding-trajectory-level-intervention-for-inference-timeSecFeedde2026-08-26T04:47:25.979460+00:00Selective Regenerative Decoding: Trajectory-Level Intervention for Inference-Time Reasoninghttps://blog.schoergendorfer.com/post/sonarllm-a-native-sonar-optical-multimodal-large-language-model-for-underwater-pSecFeedde2026-08-26T04:47:25.952086+00:00SonarLLM: A Native Sonar--Optical Multimodal Large Language Model for Underwater Perceptionhttps://blog.schoergendorfer.com/post/can-a-dynamic-internal-field-govern-a-transformers-cognition-certifiability-notSecFeedde2026-08-26T04:47:25.929137+00:00Can a Dynamic Internal Field Govern a Transformer's Cognition? Certifiability, not Superiority, in Homeostatic Compute Controlhttps://blog.schoergendorfer.com/post/benchmarking-llm-judges-for-voice-agent-evaluation-reliability-calibration-and-hSecFeedde2026-08-26T04:47:25.908162+00:00Benchmarking LLM Judges for Voice-Agent Evaluation: Reliability, Calibration, and Human Oversighthttps://blog.schoergendorfer.com/post/opdsearch-on-policy-distillation-with-rl-refinement-for-search-augmented-reasoniSecFeedde2026-08-26T04:47:25.880168+00:00OPDSearch+: On-Policy Distillation with RL Refinement for Search-Augmented Reasoninghttps://blog.schoergendorfer.com/post/videoharness-rsi-recursive-harness-self-improvement-for-long-video-understandingSecFeedde2026-08-26T04:47:25.855201+00:00VideoHarness-RSI: Recursive Harness Self-Improvement for Long-Video Understanding with Frozen Vision-Language Modelshttps://blog.schoergendorfer.com/post/reproagent-contract-guided-paper-to-code-reproductionSecFeedde2026-08-26T04:47:25.829405+00:00ReproAgent: Contract-Guided Paper-to-Code Reproductionhttps://blog.schoergendorfer.com/post/eating-for-a-sustainable-planet-personalized-sustainable-diet-recommendation-viaSecFeedde2026-08-26T04:47:25.784541+00:00Eating for a Sustainable Planet: Personalized Sustainable Diet Recommendation via Constraint-Aware Decision-Making Modelinghttps://blog.schoergendorfer.com/post/matched-excess-outranker-regularization-for-candidate-set-interference-in-continSecFeedde2026-08-26T04:47:25.753640+00:00Matched Excess-Outranker Regularization for Candidate-Set Interference in Continual Knowledge Graph Embeddinghttps://blog.schoergendorfer.com/post/real-world-knowledge-guided-change-data-synthesis-for-remote-sensingSecFeedde2026-08-26T04:47:25.723036+00:00Real-World Knowledge-Guided Change Data Synthesis for Remote Sensinghttps://blog.schoergendorfer.com/post/beyond-accuracy-a-dual-judge-evaluation-protocol-for-vision-language-models-in-lSecFeedde2026-08-26T04:47:25.697125+00:00Beyond Accuracy: A Dual-Judge Evaluation Protocol for Vision-Language Models in Legally Grounded Taskshttps://blog.schoergendorfer.com/post/sa-bench-evaluating-semantic-alignment-in-llm-based-paper-reproductionSecFeedde2026-08-26T04:47:25.671330+00:00SA-Bench: Evaluating Semantic Alignment in LLM-Based Paper Reproductionhttps://blog.schoergendorfer.com/post/strive-multi-agent-structured-temporal-reasoning-with-integrated-verification-foSecFeedde2026-08-26T04:47:25.639050+00:00STRIVE: Multi-Agent Structured Temporal Reasoning with Integrated Verification for Longitudinal Radiology Report Generationhttps://blog.schoergendorfer.com/post/trace-an-evidence-grounded-benchmark-for-safety-evaluation-of-large-reasoning-moSecFeedde2026-08-26T04:47:25.611452+00:00TRACE: An Evidence-Grounded Benchmark for Safety Evaluation of Large Reasoning Modelshttps://blog.schoergendorfer.com/post/evaluating-multiple-llm-generations-with-validated-task-coverageSecFeedde2026-08-26T04:47:25.589224+00:00Evaluating Multiple LLM Generations with Validated Task Coveragehttps://blog.schoergendorfer.com/post/constraint-guided-enterprise-data-mapping-with-large-language-modelsSecFeedde2026-08-26T04:47:25.566180+00:00Constraint-Guided Enterprise Data Mapping with Large Language Modelshttps://blog.schoergendorfer.com/post/metarag-belief-action-aligned-policy-optimization-for-agentic-ragSecFeedde2026-08-26T04:47:25.543427+00:00MetaRAG: Belief-Action Aligned Policy Optimization for Agentic RAGhttps://blog.schoergendorfer.com/post/preference-data-selection-for-mitigating-the-alignment-tax-in-large-language-modSecFeedde2026-08-26T04:47:25.514809+00:00Preference Data Selection for Mitigating the Alignment Tax in Large Language Modelshttps://blog.schoergendorfer.com/post/paritok-4b-intent-conditioned-context-compression-for-coding-agentsSecFeedde2026-08-26T04:32:47.116065+00:00Paritok-4B: Intent-Conditioned Context Compression for Coding Agentshttps://blog.schoergendorfer.com/post/task-adaptive-rubrics-for-gui-reward-modelingSecFeedde2026-08-26T04:32:47.086180+00:00Task-Adaptive Rubrics for GUI Reward Modelinghttps://blog.schoergendorfer.com/post/omnijudge-or-omnibias-diagnosing-multimodal-judges-through-balanced-decoupled-leSecFeedde2026-08-26T04:32:47.060944+00:00OmniJudge or OmniBias? Diagnosing Multimodal Judges through Balanced, Decoupled Lenseshttps://blog.schoergendorfer.com/post/robust-code-rl-via-faulty-code-driven-test-case-synthesis-and-dense-reward-shapiSecFeedde2026-08-26T04:32:47.030826+00:00Robust Code RL via Faulty-Code-Driven Test case Synthesis and Dense Reward Shapinghttps://blog.schoergendorfer.com/post/ahead-adaptive-hindsight-with-environment-augmented-distillation-for-agentic-rlSecFeedde2026-08-26T04:32:46.996093+00:00AHEAD: Adaptive Hindsight with Environment-Augmented Distillation for Agentic RLhttps://blog.schoergendorfer.com/post/scalable-question-centric-text-to-image-evaluation-reliable-ranking-fine-grainedSecFeedde2026-08-26T04:32:46.968157+00:00Scalable Question-Centric Text-to-Image Evaluation: Reliable Ranking, Fine-Grained Diagnosis, and Cost-Aware Routinghttps://blog.schoergendorfer.com/post/ace-a-self-correcting-agentic-canvas-editor-for-multi-slide-presentation-automatSecFeedde2026-08-26T04:32:46.933263+00:00ACE: A Self-Correcting Agentic Canvas Editor for Multi-Slide Presentation Automationhttps://blog.schoergendorfer.com/post/are-android-gui-agents-robust-against-runtime-anomalies-antrap-evaluating-agentsSecFeedde2026-08-26T04:32:46.903772+00:00Are Android GUI Agents Robust Against Runtime Anomalies? AnTrap: Evaluating Agents in Dynamic Adversarial Environmentshttps://blog.schoergendorfer.com/post/emrb-a-multi-level-benchmark-for-evaluating-llm-reasoning-over-raw-electromagnetSecFeedde2026-08-26T04:32:46.876138+00:00EMRB: A Multi-Level Benchmark for Evaluating LLM Reasoning over Raw Electromagnetic Signalshttps://blog.schoergendorfer.com/post/agentworld-personality-aware-reliability-evaluation-for-agentic-information-retrSecFeedde2026-08-26T04:32:46.847380+00:00AgentWorld: Personality-Aware Reliability Evaluation for Agentic Information Retrievalhttps://blog.schoergendorfer.com/post/compression-trinity-exploring-sparsity-quantization-and-low-rank-approximationsSecFeedde2026-08-26T04:32:46.815531+00:00Compression Trinity: Exploring Sparsity, Quantization, and Low-Rank Approximations for LLM Compressionhttps://blog.schoergendorfer.com/post/poisoning-agentic-alpha-adversarial-vulnerabilities-across-roles-and-architecturSecFeedde2026-08-26T04:32:46.778134+00:00Poisoning Agentic Alpha: Adversarial Vulnerabilities Across Roles and Architectures in Multi-Agent Trading Systemshttps://blog.schoergendorfer.com/post/algorithmic-impact-reveals-the-hidden-social-choice-structure-of-alignmentSecFeedde2026-08-26T04:32:46.750908+00:00Algorithmic Impact Reveals the Hidden Social Choice Structure of Alignmenthttps://blog.schoergendorfer.com/post/relative-time-intervals-representation-for-word-level-timestamping-with-masked-tSecFeedde2026-08-26T04:32:46.718461+00:00Relative Time Intervals Representation for Word-level Timestamping with Masked Traininghttps://blog.schoergendorfer.com/post/beyond-confidence-test-time-scaling-for-multi-turn-search-agents-via-retrieval-gSecFeedde2026-08-26T04:32:46.686560+00:00Beyond Confidence: Test-Time Scaling for Multi-Turn Search Agents via Retrieval Groundinghttps://blog.schoergendorfer.com/post/reflection-with-action-induced-visual-differences-for-desktop-gui-agentsSecFeedde2026-08-26T04:32:46.653313+00:00Reflection with Action-Induced Visual Differences for Desktop GUI Agentshttps://blog.schoergendorfer.com/post/incorporating-cognitive-load-and-knowledge-transfer-for-multi-domain-knowledge-tSecFeedde2026-08-26T04:32:46.625796+00:00Incorporating Cognitive Load and Knowledge Transfer for Multi-Domain Knowledge Tracinghttps://blog.schoergendorfer.com/post/diverse-by-reasoning-harnessing-the-wisdom-of-llm-crowds-for-future-predictionSecFeedde2026-08-26T04:32:46.593660+00:00Diverse by Reasoning: Harnessing the Wisdom of LLM Crowds for Future Predictionhttps://blog.schoergendorfer.com/post/memory-is-not-always-needed-characterizing-conditional-memory-in-scientific-reasSecFeedde2026-08-26T04:32:46.560128+00:00Memory Is Not Always Needed: Characterizing Conditional Memory in Scientific Reasoninghttps://blog.schoergendorfer.com/post/rules-before-oracles-auditable-user-configurable-argument-selection-for-deliberaSecFeedde2026-08-26T04:32:46.515412+00:00Rules Before Oracles: Auditable, User-Configurable Argument Selection for Deliberative Pollinghttps://blog.schoergendorfer.com/post/when-seeing-is-not-enough-benchmarking-interactive-visual-grounding-in-lvlmsSecFeedde2026-08-26T04:17:31.207631+00:00When Seeing Is Not Enough: Benchmarking Interactive Visual Grounding in LVLMshttps://blog.schoergendorfer.com/post/giraffe-a-mapping-architecture-from-hidden-text-representations-to-visual-embeddSecFeedde2026-08-26T04:17:31.182728+00:00Giraffe: A Mapping Architecture from Hidden Text Representations to Visual Embeddings for Efficient Graphic Designhttps://blog.schoergendorfer.com/post/more-gpus-or-a-smaller-cache-tensor-parallelism-versus-kv-compression-for-memorySecFeedde2026-08-26T04:17:31.158386+00:00More GPUs or a Smaller Cache? Tensor Parallelism versus KV Compression for Memory-Bound LLM Servinghttps://blog.schoergendorfer.com/post/recursive-agentic-reasoningSecFeedde2026-08-26T04:17:31.131952+00:00Recursive Agentic Reasoninghttps://blog.schoergendorfer.com/post/more-rejective-not-more-discriminative-the-unit-of-verification-in-pre-executionSecFeedde2026-08-26T04:17:31.105437+00:00More Rejective, Not More Discriminative: The Unit of Verification in Pre-Execution LLM Oversighthttps://blog.schoergendorfer.com/post/data-mixing-as-mixture-experiment-response-surface-methodology-and-optimal-desigSecFeedde2026-08-26T04:17:31.047676+00:00Data Mixing as Mixture Experiment: Response Surface Methodology and Optimal Design for Large Language Model Pretraininghttps://blog.schoergendorfer.com/post/mars-multi-specialist-llm-relay-system-for-competitive-programmingSecFeedde2026-08-26T04:17:31.010425+00:00MARS: Multi-Specialist LLM Relay System for Competitive Programminghttps://blog.schoergendorfer.com/post/proof-gen-from-optimized-data-to-better-distillationSecFeedde2026-08-26T04:17:30.977184+00:00PROOF-Gen: From Optimized Data to Better Distillationhttps://blog.schoergendorfer.com/post/retrieval-augmented-generation-vs-deterministic-tax-computation-in-multi-agent-fSecFeedde2026-08-26T04:17:30.942457+00:00Retrieval-augmented generation vs. deterministic tax computation in multi-agent financial advisory: A 2x2 factorial experimenthttps://blog.schoergendorfer.com/post/quantifying-system-level-harms-from-ai-adoption-in-complex-sociotechnical-systemSecFeedde2026-08-26T04:17:30.911434+00:00Quantifying System-Level Harms from AI Adoption in Complex Sociotechnical Systemshttps://blog.schoergendorfer.com/post/benchbench-protocol-evaluating-real-world-wet-lab-protocol-reasoning-and-modificSecFeedde2026-08-26T04:17:30.880023+00:00BenchBench-Protocol: Evaluating Real-World Wet-Lab Protocol Reasoning and Modificationhttps://blog.schoergendorfer.com/post/provenance-guided-incremental-learning-under-evolving-concept-definitionsSecFeedde2026-08-26T04:17:30.852504+00:00Provenance Guided Incremental Learning Under Evolving Concept Definitionshttps://blog.schoergendorfer.com/post/ai-finds-a-waySecFeedde2026-08-26T04:17:30.813304+00:00AI Finds A Wayhttps://blog.schoergendorfer.com/post/semantic-overlays-mitigating-prompt-injection-with-annotations-beyond-tokens-andSecFeedde2026-08-26T04:17:30.784900+00:00Semantic Overlays: Mitigating Prompt Injection with Annotations Beyond Tokens and Steering Vectorshttps://blog.schoergendorfer.com/post/granitetrust-policy-tools-shareable-actionable-policies-for-generative-ai-applicSecFeedde2026-08-26T04:17:30.756419+00:00Granite.Trust Policy Tools: Shareable, Actionable Policies for Generative AI Applicationshttps://blog.schoergendorfer.com/post/exploit-more-explore-smarter-for-budget-constrained-agentic-searchSecFeedde2026-08-26T04:17:30.702161+00:00Exploit More, Explore Smarter for Budget-Constrained Agentic Searchhttps://blog.schoergendorfer.com/post/syps-measuring-sycophancy-prompt-sensitivity-in-large-language-modelsSecFeedde2026-08-26T04:17:30.671919+00:00SyPS: Measuring Sycophancy Prompt Sensitivity in Large Language Modelshttps://blog.schoergendorfer.com/post/minima-kv-retention-preserving-kv-cache-compression-with-mixed-format-paged-atteSecFeedde2026-08-26T04:17:30.639468+00:00Minima-KV: Retention-Preserving KV Cache Compression with Mixed-Format Paged Attentionhttps://blog.schoergendorfer.com/post/a-formal-methodological-framework-for-auditing-robustness-and-fidelity-in-explaiSecFeedde2026-08-26T04:02:50.921364+00:00A Formal Methodological Framework for Auditing Robustness and Fidelity in Explainable AI: From Application to Trust Certificationhttps://blog.schoergendorfer.com/post/generating-biomedical-fact-checking-reports-with-rl-enhanced-agentic-searchSecFeedde2026-08-26T04:02:50.889259+00:00Generating Biomedical Fact-Checking Reports with RL-Enhanced Agentic Searchhttps://blog.schoergendorfer.com/post/serving-masked-diffusion-llms-characterization-and-design-principles-from-real-hSecFeedde2026-08-26T04:02:50.856346+00:00Serving Masked Diffusion LLMs: Characterization and Design Principles from Real Hardwarehttps://blog.schoergendorfer.com/post/agentroom-concurrent-multi-agent-coding-in-a-crdt-backed-shared-workspaceSecFeedde2026-08-26T04:02:50.827701+00:00AgentRoom: Concurrent Multi-Agent Coding in a CRDT-Backed Shared Workspacehttps://blog.schoergendorfer.com/post/autonomous-mathematical-discovery-in-an-open-world-multi-agent-environmentSecFeedde2026-08-26T04:02:50.774634+00:00Autonomous Mathematical Discovery in an Open-World Multi-Agent Environmenthttps://blog.schoergendorfer.com/post/automata-from-agent-traces-failure-and-next-step-predictionSecFeedde2026-08-26T04:02:50.741755+00:00Automata from Agent Traces: Failure and Next-Step Predictionhttps://blog.schoergendorfer.com/post/gated-activation-steering-for-reducing-sycophancy-hallucination-in-medical-questSecFeedde2026-08-26T04:02:50.700996+00:00Gated Activation Steering for Reducing Sycophancy & Hallucination in Medical Question Answeringhttps://blog.schoergendorfer.com/post/molemb-multimodal-large-language-models-can-be-strong-molecular-embedding-modelsSecFeedde2026-08-26T04:02:50.662295+00:00MolEmb: Multimodal Large Language Models Can Be Strong Molecular Embedding Modelshttps://blog.schoergendorfer.com/post/ethical-llm-assisted-research-a-framework-for-responsible-delegation-verificatioSecFeedde2026-08-26T04:02:50.627419+00:00Ethical LLM-Assisted Research: A Framework for Responsible Delegation, Verification, and Epistemic Valuehttps://blog.schoergendorfer.com/post/flare-a-systematic-uncertainty-aware-framework-for-evidence-based-adoption-of-arSecFeedde2026-08-26T04:02:50.592911+00:00FLARE: A Systematic, Uncertainty-Aware Framework for Evidence-Based Adoption of Artificial Intelligence in Healthcarehttps://blog.schoergendorfer.com/post/ai-agents-push-humans-out-of-the-loopSecFeedde2026-08-26T04:02:50.549603+00:00AI Agents Push Humans Out of the Loophttps://blog.schoergendorfer.com/post/how-much-of-a-measured-ai-preference-is-the-model-and-how-much-is-the-instrumentSecFeedde2026-08-26T04:02:50.518621+00:00How much of a measured AI preference is the model, and how much is the instrument?https://blog.schoergendorfer.com/post/auditing-the-synthetic-memoir-measuring-scene-level-confabulation-in-llm-generatSecFeedde2026-08-26T04:02:50.478502+00:00Auditing the Synthetic Memoir: Measuring Scene-Level Confabulation in LLM-Generated Autobiography Against the Documented Record of the Life It Describeshttps://blog.schoergendorfer.com/post/trace-transition-aware-residual-control-for-multi-objective-materials-discoverySecFeedde2026-08-26T04:02:50.416568+00:00TRACE: Transition-Aware Residual Control for Multi-Objective Materials Discoveryhttps://blog.schoergendorfer.com/post/a-survey-detection-channel-overrides-the-pixels-in-an-astronomical-foundation-moSecFeedde2026-08-26T04:02:50.388235+00:00A survey detection channel overrides the pixels in an astronomical foundation model, and biases tomographic mean redshiftshttps://blog.schoergendorfer.com/post/llm-agents-perform-controlled-experiments-using-simulation-modelsSecFeedde2026-08-26T04:02:50.349453+00:00LLM Agents Perform Controlled Experiments Using Simulation Modelshttps://blog.schoergendorfer.com/post/esq-bench-a-multi-tier-enterprise-oracle-benchmark-for-evaluating-nl2sql-dialectSecFeedde2026-08-26T04:02:50.312003+00:00ESQ-Bench: A Multi-Tier Enterprise Oracle Benchmark for Evaluating NL2SQL Dialect Generalization and Silent Semantic Divergencehttps://blog.schoergendorfer.com/post/render-controlling-reader-facing-evidence-in-llm-memory-evaluationSecFeedde2026-08-26T04:02:50.196369+00:00RENDER: Controlling Reader-Facing Evidence in LLM Memory Evaluationhttps://blog.schoergendorfer.com/post/liquid-ai-open-sources-pipette-a-reproducible-benchmarking-suite-that-measures-oSecFeedde2026-08-26T00:02:35.831169+00:00Liquid AI Open-Sources Pipette: A Reproducible Benchmarking Suite That Measures On-Device Models, Quantization, Runtime and Hardware Togetherhttps://blog.schoergendorfer.com/post/aws-lambda-microvms-now-supports-aws-privatelinkSecFeedde2026-08-25T23:17:26.171200+00:00AWS Lambda MicroVMs now supports AWS PrivateLinkhttps://blog.schoergendorfer.com/post/capacity-reservation-resource-groups-now-support-amazon-ec2-capacity-blocks-andSecFeedde2026-08-25T23:17:26.142464+00:00Capacity Reservation Resource Groups now support Amazon EC2 Capacity Blocks and interruptible Capacity Reservationshttps://blog.schoergendorfer.com/post/eve-online-the-move-to-python-3-beginsSecFeedde2026-08-25T23:17:26.108336+00:00EVE Online: The Move to Python 3 Begins!https://blog.schoergendorfer.com/post/moving-from-minimus-to-docker-hardened-imagesSecFeedde2026-08-25T23:02:37.225888+00:00Moving from Minimus to Docker Hardened Imageshttps://blog.schoergendorfer.com/post/aws-iot-core-now-supports-native-influxdb-routing-for-time-series-dataSecFeedde2026-08-25T22:17:43.577148+00:00AWS IoT Core now supports native InfluxDB routing for time-series datahttps://blog.schoergendorfer.com/post/fast-track-ism-ready-cloud-environments-and-irap-assessments-with-landing-zone-aSecFeedde2026-08-25T22:02:45.437508+00:00Fast Track ISM-ready cloud environments and IRAP Assessments with Landing Zone Accelerator on AWShttps://blog.schoergendorfer.com/post/how-to-evaluate-llms-before-productionSecFeedde2026-08-25T21:48:02.313567+00:00How to evaluate LLMs before productionhttps://blog.schoergendorfer.com/post/usn-8666-2-linux-kernel-azure-vulnerabilitiesSecFeedde2026-08-25T21:32:40.628630+00:00USN-8666-2: Linux kernel (Azure) vulnerabilitieshttps://blog.schoergendorfer.com/post/aws-lambda-introduces-managed-runtimes-in-public-preview-for-nodejs-26-and-pythoSecFeedde2026-08-25T21:17:38.172982+00:00AWS Lambda introduces managed runtimes in public preview for Node.js 26 and Python 3.15https://blog.schoergendorfer.com/post/iam-roles-anywhere-now-provides-a-java-plugin-for-the-aws-sdkSecFeedde2026-08-25T21:17:38.142859+00:00IAM Roles Anywhere now provides a Java plugin for the AWS SDKhttps://blog.schoergendorfer.com/post/is-java-enterprise-ready-for-ai-absolutelySecFeedde2026-08-25T20:31:47.319235+00:00Is Java Enterprise Ready for AI? Absolutelyhttps://blog.schoergendorfer.com/post/openais-jalapeno-chip-tackles-a-problem-ai-agents-make-worseSecFeedde2026-08-25T20:17:59.969420+00:00OpenAI’s Jalapeño chip tackles a problem AI agents make worsehttps://blog.schoergendorfer.com/post/aws-secrets-manager-adds-managed-external-secrets-support-for-cisco-security-plaSecFeedde2026-08-25T20:17:59.936658+00:00AWS Secrets Manager adds managed external secrets support for Cisco Security Platform and Netskopehttps://blog.schoergendorfer.com/post/august-25-2026SecFeedde2026-08-25T20:17:59.875336+00:00August 25, 2026https://blog.schoergendorfer.com/post/shopifys-ceo-threatened-to-ban-claude-code-anthropic-had-already-closed-the-featSecFeedde2026-08-25T19:47:24.253917+00:00Shopify’s CEO threatened to ban Claude Code. Anthropic had already closed the feature request.https://blog.schoergendorfer.com/post/the-blind-spot-between-backend-metrics-and-user-experienceSecFeedde2026-08-25T19:47:24.216368+00:00The Blind Spot Between Back‐End Metrics and User Experiencehttps://blog.schoergendorfer.com/post/cicd-for-ai-enabled-applications-why-traditional-deployment-pipelines-need-to-evSecFeedde2026-08-25T19:47:24.182456+00:00CI/CD for AI-Enabled Applications: Why Traditional Deployment Pipelines Need to Evolvehttps://blog.schoergendorfer.com/post/bendix-ec80-brake-ecuSecFeedde2026-08-25T19:47:24.141239+00:00Bendix EC80 Brake ECUhttps://blog.schoergendorfer.com/post/furuno-fa-50-class-b-ais-transponderSecFeedde2026-08-25T19:47:24.110014+00:00FURUNO FA-50 Class B AIS Transponderhttps://blog.schoergendorfer.com/post/siemens-simatic-iot2050-advancedSecFeedde2026-08-25T19:47:24.080040+00:00Siemens SIMATIC IoT2050 Advancedhttps://blog.schoergendorfer.com/post/zoneminderSecFeedde2026-08-25T19:47:24.051464+00:00Zoneminderhttps://blog.schoergendorfer.com/post/payrange-apiSecFeedde2026-08-25T19:47:24.020034+00:00PayRange APIhttps://blog.schoergendorfer.com/post/rently-smart-homeSecFeedde2026-08-25T19:47:23.986769+00:00Rently Smart Homehttps://blog.schoergendorfer.com/post/ebyte-ne2-d11SecFeedde2026-08-25T19:47:23.950129+00:00Ebyte NE2-D11https://blog.schoergendorfer.com/post/ibms-new-granite-42-models-add-reasoning-and-stay-denseSecFeedde2026-08-25T19:31:34.819990+00:00IBM’s new Granite 4.2 models add reasoning and stay densehttps://blog.schoergendorfer.com/post/the-grafana-ai-sdk-for-go-a-shared-foundation-for-building-ai-applicationsSecFeedde2026-08-25T19:31:34.790797+00:00The Grafana AI SDK for Go: a shared foundation for building AI applicationshttps://blog.schoergendorfer.com/post/how-telemetry-pipelines-keep-ai-agent-costs-under-controlSecFeedde2026-08-25T19:16:37.235770+00:00How telemetry pipelines keep AI agent costs under controlhttps://blog.schoergendorfer.com/post/why-dns-dhcp-and-ipam-can-no-longer-live-in-separate-silosSecFeedde2026-08-25T19:16:37.207069+00:00Why DNS, DHCP, and IPAM Can No Longer Live in Separate Siloshttps://blog.schoergendorfer.com/post/amazon-rds-for-oracle-now-supports-july-2026-release-updateSecFeedde2026-08-25T19:16:37.182519+00:00Amazon RDS for Oracle now supports July 2026 Release Updatehttps://blog.schoergendorfer.com/post/usn-8678-1-openssl-vulnerabilitiesSecFeedde2026-08-25T19:16:37.159115+00:00USN-8678-1: OpenSSL vulnerabilitieshttps://blog.schoergendorfer.com/post/perplexity-ships-portable-computer-on-nvidia-dgx-spark-local-harness-os-enforcedSecFeedde2026-08-25T19:16:37.125736+00:00Perplexity Ships Portable Computer on NVIDIA DGX Spark: Local Harness, OS-Enforced Sandbox, and Zero Per-Token Cost for Local Stepshttps://blog.schoergendorfer.com/post/agentic-observability-with-amazon-opensearch-service-mcp-appsSecFeedde2026-08-25T19:02:08.655562+00:00Agentic observability with Amazon OpenSearch Service MCP Appshttps://blog.schoergendorfer.com/post/cisa-adds-one-known-exploited-vulnerability-to-catalog-13SecFeedde2026-08-25T18:46:38.049550+00:00CISA Adds One Known Exploited Vulnerability to Cataloghttps://blog.schoergendorfer.com/post/cli-or-ide-build-in-verification-first-2SecFeedde2026-08-25T18:31:47.480225+00:00CLI or IDE? Build in verification first.https://blog.schoergendorfer.com/post/msk-replicator-now-supports-oauth-20-sasloauthbearer-authentication-for-replicatSecFeedde2026-08-25T18:17:37.765102+00:00MSK Replicator now supports OAuth 2.0 (SASL/OAUTHBEARER) authentication for replication from external Apache Kafka clusters to Amazon MSKhttps://blog.schoergendorfer.com/post/amazon-rds-for-postgresql-supports-minor-versions-186-1711-1615-1519-and-1424SecFeedde2026-08-25T18:17:37.734928+00:00Amazon RDS for PostgreSQL supports minor versions 18.6, 17.11, 16.15, 15.19, and 14.24https://blog.schoergendorfer.com/post/meta-ai-introduces-metaroce-a-clean-sheet-rdma-transport-built-for-ai-scale-etheSecFeedde2026-08-25T17:32:07.595056+00:00Meta AI Introduces MetaRoCE: A Clean-Sheet RDMA Transport Built for AI-Scale Ethernethttps://blog.schoergendorfer.com/post/anthropic-gives-chat-and-cowork-one-memorySecFeedde2026-08-25T17:16:37.243363+00:00Anthropic gives chat and Cowork one memoryhttps://blog.schoergendorfer.com/post/governed-reports-with-amazon-quick-desktop-and-amazon-fsx-for-netapp-ontapSecFeedde2026-08-25T17:16:37.161627+00:00Governed reports with Amazon Quick Desktop and Amazon FSx for NetApp ONTAPhttps://blog.schoergendorfer.com/post/the-patch-window-is-collapsing-why-security-needs-a-new-control-planeSecFeedde2026-08-25T17:16:37.064258+00:00The patch window is collapsing: Why security needs a new control planehttps://blog.schoergendorfer.com/post/the-ai-agent-race-is-on-but-are-we-watching-the-right-raceSecFeedde2026-08-25T17:01:42.996381+00:00The AI Agent Race Is On. But Are We Watching the Right Race?https://blog.schoergendorfer.com/post/happy-20th-birthday-amazon-ec2SecFeedde2026-08-25T17:01:42.964552+00:00Happy 20th Birthday, Amazon EC2https://blog.schoergendorfer.com/post/bringing-gvisor-sandboxes-to-distributed-ray-clustersSecFeedde2026-08-25T17:01:42.930684+00:00Bringing gVisor sandboxes to distributed Ray clustershttps://blog.schoergendorfer.com/post/tokenmaxxing-is-out-how-to-minimize-ai-spend-without-sacrificing-security-capabiSecFeedde2026-08-25T17:01:42.899606+00:00Tokenmaxxing is out. How to minimize AI spend without sacrificing security capability.https://blog.schoergendorfer.com/post/perplexitys-computer-agent-can-now-run-locally-if-you-can-afford-itSecFeedde2026-08-25T17:01:41.785385+00:00Perplexity’s Computer agent can now run locally — if you can afford ithttps://blog.schoergendorfer.com/post/obfuscating-ip-addresses-as-hostnames-tue-aug-25thSecFeedde2026-08-25T16:46:36.666263+00:00Obfuscating IP Addresses as Hostnames, (Tue, Aug 25th)https://blog.schoergendorfer.com/post/a-tale-of-two-socs-insights-from-two-red-team-assessmentsSecFeedde2026-08-25T16:31:42.954207+00:00A Tale of Two SOCs: Insights From Two Red Team Assessmentshttps://blog.schoergendorfer.com/post/cli-or-ide-build-in-verification-firstSecFeedde2026-08-25T16:03:54.345046+00:00CLI or IDE? Build in verification firsthttps://blog.schoergendorfer.com/post/openai-built-a-chip-in-nine-months-then-it-let-ai-rewrite-the-codeSecFeedde2026-08-25T16:03:54.316661+00:00OpenAI built a chip in nine months. Then it let AI rewrite the code.https://blog.schoergendorfer.com/post/last-week-in-ai-342-last-3-months-in-aiSecFeedde2026-08-25T15:48:21.816464+00:00Last Week in AI #342 - Last 3 Months in AIhttps://blog.schoergendorfer.com/post/you-can-rent-a-feature-but-you-cant-rent-a-foundation-why-motherduck-bought-theSecFeedde2026-08-25T15:48:21.779327+00:00“You can rent a feature, but you can’t rent a foundation”: why MotherDuck bought the startup already powering its data pipelineshttps://blog.schoergendorfer.com/post/ai-agents-are-spreading-fast-their-rules-are-still-catching-upSecFeedde2026-08-25T15:48:21.657654+00:00AI agents are spreading fast. Their rules are still catching up.https://blog.schoergendorfer.com/post/report-shines-spotlight-on-91-vulnerabilities-fixed-in-latest-update-to-spring-fSecFeedde2026-08-25T15:35:24.927156+00:00Report Shines Spotlight on 91 Vulnerabilities Fixed in Latest Update to Spring Frameworkhttps://blog.schoergendorfer.com/post/now-introducing-gemini-enterprise-for-financial-servicesSecFeedde2026-08-25T15:35:23.909627+00:00Now introducing Gemini Enterprise for Financial Serviceshttps://blog.schoergendorfer.com/post/now-introducing-gemini-enterprise-for-legalSecFeedde2026-08-25T15:35:23.870218+00:00Now introducing Gemini Enterprise for Legalhttps://blog.schoergendorfer.com/post/scale-software-delivery-pipelines-in-isolation-without-owning-the-runner-fleetSecFeedde2026-08-25T15:17:42.833371+00:00Scale software delivery pipelines in isolation without owning the runner fleethttps://blog.schoergendorfer.com/post/the-lazy-developers-guide-to-observing-your-own-codeSecFeedde2026-08-25T15:17:42.804938+00:00The lazy developer’s guide to observing your own codehttps://blog.schoergendorfer.com/post/stop-trying-to-learn-all-of-kubernetes-at-onceSecFeedde2026-08-25T15:03:24.912158+00:00Stop trying to learn all of Kubernetes at oncehttps://blog.schoergendorfer.com/post/ohne-titel-7SecFeedde2026-08-25T15:03:23.909524+00:00Ohne Titelhttps://blog.schoergendorfer.com/post/aws-lambda-functions-now-support-full-iam-resource-based-policiesSecFeedde2026-08-25T14:48:49.855614+00:00AWS Lambda functions now support full IAM resource-based policieshttps://blog.schoergendorfer.com/post/the-safety-penalty-reclaiming-operational-sovereignty-in-the-age-of-aiSecFeedde2026-08-25T14:34:19.727803+00:00The safety penalty: Reclaiming operational sovereignty in the age of AIhttps://blog.schoergendorfer.com/post/i-spent-a-day-at-a-robot-carnival-in-shanghai-heres-what-i-sawSecFeedde2026-08-25T14:34:19.613563+00:00I spent a day at a robot “carnival” in Shanghai. Here’s what I saw.https://blog.schoergendorfer.com/post/multi-agent-orchestration-with-the-common-sense-reasoning-capabilities-of-llms-f-2SecFeedde2026-08-25T13:34:29.431660+00:00Multi-Agent Orchestration with the Common-Sense Reasoning Capabilities of LLMs for Autonomous Drivinghttps://blog.schoergendorfer.com/post/decomposition-attacks-across-unlinkable-identities-limits-of-stateful-defenses-f-2SecFeedde2026-08-25T13:34:29.396562+00:00Decomposition Attacks Across Unlinkable Identities: Limits of Stateful Defenses for LLM Serviceshttps://blog.schoergendorfer.com/post/towards-understanding-on-policy-distillation-through-the-lens-of-test-time-scali-2SecFeedde2026-08-25T13:34:29.357711+00:00Towards Understanding On-Policy Distillation through the Lens of Test-Time Scalinghttps://blog.schoergendorfer.com/post/do-llm-recommenders-know-when-theyre-hallucinating-auditing-confidence-calibrati-3SecFeedde2026-08-25T13:34:29.315189+00:00Do LLM Recommenders Know When They're Hallucinating? Auditing Confidence Calibration in Catalog Faithfulnesshttps://blog.schoergendorfer.com/post/macaron-v1-towards-open-continual-learning-with-self-improvement-and-mixture-of-2SecFeedde2026-08-25T13:16:32.620422+00:00Macaron-V1: Towards Open Continual Learning with Self-Improvement and Mixture-of-LoRAhttps://blog.schoergendorfer.com/post/a-jolt-for-the-kv-cache-near-lossless-kv-cache-compression-via-joint-lagrangianSecFeedde2026-08-25T13:16:32.587577+00:00A JoLT for the KV cache: Near-lossless KV cache compression via joint Lagrangian allocation of Tucker ranks and a rotated residual for llmshttps://blog.schoergendorfer.com/post/scar-semantic-continuity-aware-retrieval-for-efficient-context-expansion-in-ragSecFeedde2026-08-25T13:16:32.555313+00:00SCAR: Semantic Continuity-Aware Retrieval for Efficient Context Expansion in RAGhttps://blog.schoergendorfer.com/post/same-payload-different-channel-measuring-trust-asymmetry-in-tool-using-languageSecFeedde2026-08-25T13:16:32.497694+00:00Same Payload, Different Channel: Measuring Trust Asymmetry in Tool-Using Language Modelshttps://blog.schoergendorfer.com/post/safelens-deliberate-and-efficient-video-guardrails-with-fast-and-slow-screeningSecFeedde2026-08-25T13:16:32.463035+00:00SafeLens: Deliberate and Efficient Video Guardrails with Fast-and-Slow Screeninghttps://blog.schoergendorfer.com/post/detecting-and-suppressing-reward-hacking-with-gradient-fingerprintsSecFeedde2026-08-25T13:16:32.430647+00:00Detecting and Suppressing Reward Hacking with Gradient Fingerprintshttps://blog.schoergendorfer.com/post/scaling-electronic-health-record-foundation-models-for-population-health-managemSecFeedde2026-08-25T13:16:32.311367+00:00Scaling Electronic Health Record Foundation Models for Population Health Managementhttps://blog.schoergendorfer.com/post/forgetting-to-forget-attention-sink-as-a-gateway-for-backdooring-llm-unlearningSecFeedde2026-08-25T13:16:32.278192+00:00Forgetting to Forget: Attention Sink as A Gateway for Backdooring LLM Unlearninghttps://blog.schoergendorfer.com/post/benchmarking-and-boosting-multilingual-capabilities-of-lvlms-via-ocr-centric-reiSecFeedde2026-08-25T13:16:32.247542+00:00Benchmarking and Boosting Multilingual Capabilities of LVLMs via OCR-Centric Reinforcement Learninghttps://blog.schoergendorfer.com/post/foredreamer-a-self-evolving-dual-agent-memory-architecture-for-future-event-pred-2SecFeedde2026-08-25T13:16:32.219921+00:00ForeDreamer: A Self-Evolving Dual-Agent Memory Architecture for Future Event Predictionhttps://blog.schoergendorfer.com/post/synflow-a-multidimensional-diachronic-semantic-analysis-toolkit-2SecFeedde2026-08-25T13:16:32.190086+00:00SynFlow: A Multidimensional Diachronic Semantic Analysis Toolkithttps://blog.schoergendorfer.com/post/tracesql-traceable-answerability-estimation-for-reference-free-text-to-sql-verif-2SecFeedde2026-08-25T13:16:32.161627+00:00TraceSQL: Traceable Answerability Estimation for Reference-Free Text-to-SQL Verificationhttps://blog.schoergendorfer.com/post/every-coin-has-two-sides-on-the-dual-nature-of-generalization-in-on-policy-disti-2SecFeedde2026-08-25T13:16:32.134015+00:00Every Coin Has Two Sides: On the Dual Nature of Generalization in On-Policy Distillation of Large Language Modelshttps://blog.schoergendorfer.com/post/ask-condition-or-abstain-reinforcement-learning-for-missing-premise-reasoning-2SecFeedde2026-08-25T13:16:32.102384+00:00Ask, Condition or Abstain: Reinforcement Learning for Missing-Premise Reasoninghttps://blog.schoergendorfer.com/post/agentic-duplexgen-decoupling-content-timing-and-acoustics-for-synthetic-dialogueSecFeedde2026-08-25T13:16:32.070635+00:00Agentic-DuplexGen: Decoupling Content, Timing, and Acoustics for Synthetic Dialogue Speechhttps://blog.schoergendorfer.com/post/training-agents-to-evolve-with-their-harness-taolive-digital-avatar-agent-techniSecFeedde2026-08-25T13:16:32.041256+00:00Training Agents to Evolve with Their Harness: TaoLive Digital Avatar Agent Technical Reporthttps://blog.schoergendorfer.com/post/massive-activations-in-hybrid-linear-attention-large-language-models-pre-attenti-2SecFeedde2026-08-25T13:01:56.773836+00:00Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateaushttps://blog.schoergendorfer.com/post/sag-sql-retrieval-augmented-generation-with-query-time-dynamic-hyperedges-2SecFeedde2026-08-25T13:01:56.746206+00:00SAG: SQL-Retrieval Augmented Generation with Query-Time Dynamic Hyperedgeshttps://blog.schoergendorfer.com/post/every-token-counts-exact-likert-scale-distributions-for-measuring-llm-attitudes-2SecFeedde2026-08-25T13:01:56.718154+00:00Every Token Counts: Exact Likert-Scale Distributions for Measuring LLM Attitudes and Biaseshttps://blog.schoergendorfer.com/post/accurate-but-natural-diagnosing-grammatical-and-idiomatic-gaps-in-japanese-efl-w-2SecFeedde2026-08-25T13:01:56.689307+00:00Accurate but Natural? Diagnosing Grammatical and Idiomatic Gaps in Japanese EFL Writinghttps://blog.schoergendorfer.com/post/constructcie-a-dataset-for-extracting-causal-information-from-construction-accid-2SecFeedde2026-08-25T13:01:56.652697+00:00ConstructCIE: A Dataset for Extracting Causal Information from Construction Accident Narrativeshttps://blog.schoergendorfer.com/post/sparse-ppmi-graph-averaging-for-random-indexing-embeddingsSecFeedde2026-08-25T13:01:56.621142+00:00Sparse PPMI Graph Averaging for Random Indexing Embeddingshttps://blog.schoergendorfer.com/post/predicting-multilingual-classification-and-translation-performance-of-llms-with-2SecFeedde2026-08-25T13:01:56.595851+00:00Predicting Multilingual Classification and Translation Performance of LLMs with Cross-Lingual Alignment -- Is English Enough?https://blog.schoergendorfer.com/post/rig-rope-relation-stratified-multimodal-attention-with-instance-local-rotary-geo-2SecFeedde2026-08-25T13:01:56.565531+00:00RIG-RoPE: Relation-Stratified Multimodal Attention with Instance-Local Rotary Geometry and Representation-Aware Traversal Coordinateshttps://blog.schoergendorfer.com/post/prompt-induced-waste-in-coding-agents-reasoning-effort-harness-design-and-end-to-2SecFeedde2026-08-25T13:01:56.537843+00:00Prompt-Induced Waste in Coding Agents: Reasoning, Effort, Harness Design, and End-to-End Costhttps://blog.schoergendorfer.com/post/two-regimes-of-chain-of-thought-unfaithfulness-metric-based-detection-fails-wherSecFeedde2026-08-25T13:01:56.505701+00:00Two Regimes of Chain-of-Thought Unfaithfulness: Metric-Based Detection Fails Where Models Are Wronghttps://blog.schoergendorfer.com/post/dynakrag-a-unified-framework-for-learnable-evidence-control-in-multi-hop-retriev-2SecFeedde2026-08-25T13:01:56.471774+00:00DynaKRAG: A Unified Framework for Learnable Evidence Control in Multi-Hop Retrieval-Augmented Generationhttps://blog.schoergendorfer.com/post/how-does-research-evolve-tracing-cross-domain-trajectories-in-nlp-ml-and-cv-throSecFeedde2026-08-25T13:01:56.440737+00:00How Does Research Evolve? Tracing Cross-Domain Trajectories in NLP, ML, and CV Through Claim-Grounded Typed Citationshttps://blog.schoergendorfer.com/post/grace-step-level-benchmark-for-faithful-reasoning-over-contextSecFeedde2026-08-25T13:01:56.415523+00:00GRACE: Step-Level Benchmark for Faithful Reasoning over Contexthttps://blog.schoergendorfer.com/post/clinically-grounded-privacy-evaluation-of-medical-lmsSecFeedde2026-08-25T13:01:56.391488+00:00Clinically Grounded Privacy Evaluation of Medical LMshttps://blog.schoergendorfer.com/post/do-value-vectors-in-deep-layers-need-context-from-the-residual-stream-3SecFeedde2026-08-25T13:01:56.341099+00:00Do Value Vectors in Deep Layers Need Context from the Residual Stream?https://blog.schoergendorfer.com/post/raset-router-agnostic-safety-critical-expert-tuning-exposes-localized-safety-enfSecFeedde2026-08-25T13:01:56.310350+00:00RASET: Router-Agnostic Safety-Critical Expert Tuning Exposes Localized Safety Enforcement Failures in Mixture-of-Experts LLMshttps://blog.schoergendorfer.com/post/checkup2action-a-multimodal-clinical-check-up-report-dataset-for-patient-oriente-2SecFeedde2026-08-25T13:01:56.283600+00:00Checkup2Action: A Multimodal Clinical Check-up Report Dataset for Patient-Oriented Action Card Generationhttps://blog.schoergendorfer.com/post/unleashing-implicit-rewards-prefix-value-learning-for-distribution-level-optimiz-2SecFeedde2026-08-25T13:01:56.226020+00:00Unleashing Implicit Rewards: Prefix-Value Learning for Distribution-Level Optimizationhttps://blog.schoergendorfer.com/post/dynhd-hallucination-detection-for-diffusion-large-language-models-via-denoisingSecFeedde2026-08-25T12:46:29.823597+00:00DynHD: Hallucination Detection for Diffusion Large Language Models via Denoising Dynamics Deviation Learninghttps://blog.schoergendorfer.com/post/dialects-of-translationese-shape-language-model-learningSecFeedde2026-08-25T12:46:29.795554+00:00Dialects of Translationese Shape Language Model Learninghttps://blog.schoergendorfer.com/post/adaptive-test-time-compute-allocation-for-block-diffusion-language-models-in-comSecFeedde2026-08-25T12:46:29.762622+00:00Adaptive Test-Time Compute Allocation for Block Diffusion Language Models in Complex Reasoninghttps://blog.schoergendorfer.com/post/lakehopper-knowledge-aware-adaptation-of-column-type-annotators-across-data-lakeSecFeedde2026-08-25T12:46:29.733381+00:00LakeHopper: Knowledge-Aware Adaptation of Column Type Annotators across Data Lakeshttps://blog.schoergendorfer.com/post/mirror-a-multi-agent-framework-with-iterative-adaptive-revision-and-hierarchicalSecFeedde2026-08-25T12:46:29.703567+00:00MIRROR: A Multi-Agent Framework with Iterative Adaptive Revision and Hierarchical Retrieval for Optimization Modeling in Operations Researchhttps://blog.schoergendorfer.com/post/caliburn-self-calibrated-llm-unlearning-alignmentSecFeedde2026-08-25T12:46:29.669837+00:00CALIBURN: Self-Calibrated LLM Unlearning Alignmenthttps://blog.schoergendorfer.com/post/lookahead-then-verify-reliable-constrained-decoding-for-diffusion-llms-under-conSecFeedde2026-08-25T12:46:29.636097+00:00Lookahead-then-Verify: Reliable Constrained Decoding for Diffusion LLMs under Context-Free Grammarshttps://blog.schoergendorfer.com/post/what-language-models-know-but-dont-say-non-generative-prior-extraction-for-generSecFeedde2026-08-25T12:46:29.586753+00:00What Language Models Know But Don't Say: Non-Generative Prior Extraction for Generalizationhttps://blog.schoergendorfer.com/post/opportunities-and-challenges-of-natural-language-processing-for-low-resource-senSecFeedde2026-08-25T12:46:29.556404+00:00Opportunities and Challenges of Natural Language Processing for Low-Resource Senegalese Languages in Social Science Researchhttps://blog.schoergendorfer.com/post/luxit-a-luxembourgish-instruction-tuning-dataset-from-monolingual-seed-dataSecFeedde2026-08-25T12:46:29.530929+00:00LuxIT: A Luxembourgish Instruction Tuning Dataset from Monolingual Seed Datahttps://blog.schoergendorfer.com/post/syntax-guided-diffusion-language-models-with-user-integrated-personalizationSecFeedde2026-08-25T12:46:29.499743+00:00Syntax-Guided Diffusion Language Models with User-Integrated Personalizationhttps://blog.schoergendorfer.com/post/convergewriter-data-driven-bottom-up-article-constructionSecFeedde2026-08-25T12:46:29.469894+00:00ConvergeWriter: Data-Driven Bottom-Up Article Constructionhttps://blog.schoergendorfer.com/post/models-in-the-same-family-are-not-trust-equivalentSecFeedde2026-08-25T12:46:29.443193+00:00Models in the Same Family are NOT Trust-Equivalenthttps://blog.schoergendorfer.com/post/omni-safetybench-a-benchmark-for-safety-evaluation-of-audio-visual-large-languagSecFeedde2026-08-25T12:46:29.415661+00:00Omni-SafetyBench: A Benchmark for Safety Evaluation of Audio-Visual Large Language Modelshttps://blog.schoergendorfer.com/post/model-directions-not-words-mechanistic-topic-models-using-sparse-autoencodersSecFeedde2026-08-25T12:46:29.388096+00:00Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencodershttps://blog.schoergendorfer.com/post/safety-aligned-weights-are-not-enough-refusal-teacher-guided-finetuning-enhancesSecFeedde2026-08-25T12:46:29.361744+00:00Safety-Aligned Weights Are Not Enough: Refusal-Teacher-Guided Finetuning Enhances Safety and Downstream Performance under Harmful Finetuning Attackshttps://blog.schoergendorfer.com/post/reasoning-meets-personalization-unleashing-the-potential-of-large-reasoning-modeSecFeedde2026-08-25T12:46:29.335562+00:00Reasoning Meets Personalization: Unleashing the Potential of Large Reasoning Model for Personalized Generationhttps://blog.schoergendorfer.com/post/length-controlled-margin-based-preference-optimization-without-reference-modelSecFeedde2026-08-25T12:31:33.547687+00:00Length-Controlled Margin-Based Preference Optimization without Reference Modelhttps://blog.schoergendorfer.com/post/training-large-language-models-to-reason-in-a-continuous-latent-spaceSecFeedde2026-08-25T12:31:33.522673+00:00Training Large Language Models to Reason in a Continuous Latent Spacehttps://blog.schoergendorfer.com/post/bridging-linguistic-structure-and-mechanistic-interpretability-for-conceptual-inSecFeedde2026-08-25T12:31:33.494228+00:00Bridging Linguistic Structure and Mechanistic Interpretability for Conceptual Interpretation in Language Modelshttps://blog.schoergendorfer.com/post/robustness-of-ir-models-to-collection-growthSecFeedde2026-08-25T12:31:33.465484+00:00Robustness of IR Models to Collection Growthhttps://blog.schoergendorfer.com/post/expectations-and-practices-around-ai-disclosure-in-cs-researchSecFeedde2026-08-25T12:31:33.438140+00:00Expectations and Practices around AI Disclosure in CS Researchhttps://blog.schoergendorfer.com/post/the-laws-of-context-allocation-causal-measurement-and-closed-loop-orchestrationSecFeedde2026-08-25T12:31:33.410145+00:00The Laws of Context Allocation: Causal Measurement and Closed-Loop Orchestration in Generative Searchhttps://blog.schoergendorfer.com/post/cyberfactory-scaling-cyber-security-capabilities-with-instances-from-the-wildSecFeedde2026-08-25T12:31:33.377534+00:00CyberFactory: Scaling Cyber Security Capabilities with Instances from the Wildhttps://blog.schoergendorfer.com/post/st2u-stateful-test-time-unlearning-via-restricted-knowledge-boundary-controlSecFeedde2026-08-25T12:31:33.321895+00:00ST$^2$U: Stateful Test-Time Unlearning via Restricted Knowledge Boundary Controlhttps://blog.schoergendorfer.com/post/warp-wasserstein-aligned-rag-for-population-opinionsSecFeedde2026-08-25T12:31:33.289753+00:00WARP: Wasserstein-Aligned RAG for Population Opinionshttps://blog.schoergendorfer.com/post/same-agent-different-answers-a-repeat-aware-audit-of-corpus-induced-answer-churnSecFeedde2026-08-25T12:31:33.263952+00:00Same Agent, Different Answers: A Repeat-Aware Audit of Corpus-Induced Answer Churn in Retrieval-Augmented QAhttps://blog.schoergendorfer.com/post/trace-temporal-retrieval-with-anchored-and-convergent-evidence-for-long-horizonSecFeedde2026-08-25T12:31:33.238867+00:00TRACE: Temporal Retrieval with Anchored and Convergent Evidence for Long-Horizon Video Understandinghttps://blog.schoergendorfer.com/post/all-four-leading-llms-talk-more-than-they-listen-to-personality-verified-synthetSecFeedde2026-08-25T12:31:33.213353+00:00All four leading LLMs talk more than they listen to personality-verified synthetic help-seekershttps://blog.schoergendorfer.com/post/graft-graph-distilled-generative-retrieval-for-facet-aware-scientific-literatureSecFeedde2026-08-25T12:31:33.183904+00:00GRAFT: Graph-Distilled Generative Retrieval for Facet-Aware Scientific Literature Explorationhttps://blog.schoergendorfer.com/post/unveiling-the-depth-performance-dilemma-in-split-federated-fine-tuning-of-llmsSecFeedde2026-08-25T12:31:33.100413+00:00Unveiling the Depth-Performance Dilemma in Split-Federated Fine-tuning of LLMshttps://blog.schoergendorfer.com/post/decoupled-physical-modeling-and-execution-for-physics-reasoningSecFeedde2026-08-25T12:31:33.028173+00:00Decoupled Physical Modeling and Execution for Physics Reasoninghttps://blog.schoergendorfer.com/post/tango-token-aggregated-nonlinear-gating-operators-for-natural-and-formal-languagSecFeedde2026-08-25T12:16:33.519099+00:00TANGO: Token-Aggregated Nonlinear Gating Operators for Natural and Formal Language Modelinghttps://blog.schoergendorfer.com/post/w-rag-source-aware-retrieval-for-enterprise-document-generation-from-heterogeneoSecFeedde2026-08-25T12:16:33.491083+00:00W-RAG: Source-Aware Retrieval for Enterprise Document Generation from Heterogeneous Knowledge Baseshttps://blog.schoergendorfer.com/post/the-communication-map-of-a-transformerSecFeedde2026-08-25T12:16:33.462794+00:00The Communication Map of a Transformerhttps://blog.schoergendorfer.com/post/skillbloat-token-amplification-attacks-via-skill-injection-in-llm-coding-agentsSecFeedde2026-08-25T12:16:33.433702+00:00SkillBloat: Token Amplification Attacks via Skill Injection in LLM Coding Agentshttps://blog.schoergendorfer.com/post/learning-to-look-again-loss-gap-supervision-for-free-form-crop-routing-in-visionSecFeedde2026-08-25T12:16:33.407687+00:00Learning to Look Again: Loss-Gap Supervision for Free-form Crop Routing in Vision-Language Modelshttps://blog.schoergendorfer.com/post/the-plan-not-the-decoder-diagnosing-and-repairing-compositional-failure-in-reasoSecFeedde2026-08-25T12:16:33.378540+00:00The Plan, Not the Decoder: Diagnosing and Repairing Compositional Failure in Reasoning-Augmented Text-to-Image Generationhttps://blog.schoergendorfer.com/post/beyond-sparse-weights-when-is-attention-compressibleSecFeedde2026-08-25T12:16:33.346931+00:00Beyond Sparse Weights: When Is Attention Compressible?https://blog.schoergendorfer.com/post/damagescope-vision-language-retrieval-at-scale-for-disaster-damage-assessment-frSecFeedde2026-08-25T12:16:33.318840+00:00DamageScope: Vision-Language Retrieval at Scale for Disaster Damage Assessment from Satellite Imageryhttps://blog.schoergendorfer.com/post/generative-gap-fillingSecFeedde2026-08-25T12:16:33.293411+00:00Generative Gap Fillinghttps://blog.schoergendorfer.com/post/personamem-v3-toward-omni-platform-personal-intelligence-for-holistic-user-underSecFeedde2026-08-25T12:16:33.268826+00:00PersonaMem-v3: Toward Omni-Platform Personal Intelligence for Holistic User Understanding, Recommendation, and Agentic Taskshttps://blog.schoergendorfer.com/post/on-the-threat-model-of-weird-generalization-and-emergent-misalignmentSecFeedde2026-08-25T12:16:33.237065+00:00On the Threat Model of Weird Generalization and Emergent Misalignmenthttps://blog.schoergendorfer.com/post/stonic-a-layered-measurement-contract-for-llm-value-profilingSecFeedde2026-08-25T12:16:33.185247+00:00STONIC: A Layered Measurement Contract for LLM Value Profilinghttps://blog.schoergendorfer.com/post/a-comprehensive-analysis-of-arabic-natural-language-processing-research-trends-tSecFeedde2026-08-25T12:16:33.155523+00:00A Comprehensive Analysis of Arabic Natural Language Processing Research: Trends, Topic Evolution, and Research Gaps -- A Bibliometric and Topic-Based Studyhttps://blog.schoergendorfer.com/post/the-geometry-of-low-resource-language-representationsSecFeedde2026-08-25T12:16:33.124806+00:00The Geometry of Low-Resource Language Representationshttps://blog.schoergendorfer.com/post/formuevo-llm-guided-evolution-for-discovering-solver-efficient-mixed-integer-proSecFeedde2026-08-25T12:16:33.092802+00:00FormuEvo: LLM-Guided Evolution for Discovering Solver-Efficient Mixed-Integer Programming Formulationshttps://blog.schoergendorfer.com/post/flesch-kincaid-readability-depends-only-on-the-topic-distribution-in-long-textsSecFeedde2026-08-25T12:16:33.063346+00:00Flesch-Kincaid Readability Depends Only on the Topic Distribution in Long Texts under Topic Modelshttps://blog.schoergendorfer.com/post/beyond-the-stability-exploration-dilemma-environmental-regularization-for-llm-poSecFeedde2026-08-25T12:16:33.033184+00:00Beyond the Stability-Exploration Dilemma: Environmental Regularization for LLM Policy Optimizationhttps://blog.schoergendorfer.com/post/dynamic-topic-modeling-for-cross-corpus-temporal-analysisSecFeedde2026-08-25T12:16:33.003623+00:00Dynamic Topic Modeling for Cross-Corpus Temporal Analysishttps://blog.schoergendorfer.com/post/evowiki-incremental-state-overwriting-and-traceable-question-answering-for-crossSecFeedde2026-08-25T12:16:32.977661+00:00EvoWiki: Incremental State Overwriting and Traceable Question Answering for Cross-Meeting Knowledge Evolutionhttps://blog.schoergendorfer.com/post/a-multi-domain-and-multi-task-generative-framework-with-explicit-task-and-domainSecFeedde2026-08-25T12:02:00.601307+00:00A Multi-Domain and Multi-Task Generative Framework with Explicit Task and Domain Conditioning for Cross-Domain Event Extractionhttps://blog.schoergendorfer.com/post/aligning-biomedical-texts-and-knowledge-graphs-a-systematic-comparison-of-lightwSecFeedde2026-08-25T12:02:00.565197+00:00Aligning Biomedical Texts and Knowledge Graphs: A Systematic Comparison of Lightweight Alignment Strategieshttps://blog.schoergendorfer.com/post/longwof-bench-evaluating-evomap-genes-for-verifiable-long-workflow-tasksSecFeedde2026-08-25T12:02:00.516040+00:00LongWoF-Bench: Evaluating EvoMap Genes for Verifiable Long-Workflow Taskshttps://blog.schoergendorfer.com/post/cargo-t-causal-reasoning-graph-of-thought-improves-multimodal-humor-comprehensioSecFeedde2026-08-25T12:02:00.477484+00:00CaRGo-T: Causal Reasoning Graph-of-Thought improves Multimodal Humor Comprehensionhttps://blog.schoergendorfer.com/post/accelerating-diffusion-language-models-via-structured-suffix-modelingSecFeedde2026-08-25T12:02:00.443387+00:00Accelerating Diffusion Language Models via Structured Suffix Modelinghttps://blog.schoergendorfer.com/post/counter-with-evidence-a-multi-agent-memory-efficient-reasoning-framework-for-hatSecFeedde2026-08-25T12:02:00.413697+00:00Counter with Evidence! A Multi-Agent Memory Efficient Reasoning Framework for Hate Category Informed Counterspeech Generationhttps://blog.schoergendorfer.com/post/definitional-sensitivity-in-media-bias-detection-a-multi-definition-dataset-andSecFeedde2026-08-25T12:02:00.378003+00:00Definitional Sensitivity in Media Bias Detection: A Multi-Definition Dataset and Benchmarkhttps://blog.schoergendorfer.com/post/signal-or-noise-a-benchmark-study-of-agent-skills-in-web-developmentSecFeedde2026-08-25T12:02:00.347367+00:00Signal or Noise? A Benchmark Study of Agent Skills in Web Developmenthttps://blog.schoergendorfer.com/post/most-of-the-llm-routing-gap-is-task-typeSecFeedde2026-08-25T12:02:00.281903+00:00Most of the LLM routing gap is task typehttps://blog.schoergendorfer.com/post/llm-pedagogical-behavior-in-ai-tutoring-interactionsSecFeedde2026-08-25T12:02:00.248031+00:00LLM Pedagogical Behavior in AI Tutoring Interactionshttps://blog.schoergendorfer.com/post/what-does-activation-steering-control-attribution-across-answer-encodings-and-ouSecFeedde2026-08-25T12:02:00.215776+00:00What Does Activation Steering Control? Attribution Across Answer Encodings and Output-Sensitive Subspaceshttps://blog.schoergendorfer.com/post/closed-loop-bayesian-molecular-inverse-design-with-semantic-llm-surrogatesSecFeedde2026-08-25T12:02:00.180506+00:00Closed-Loop Bayesian Molecular Inverse Design with Semantic LLM Surrogateshttps://blog.schoergendorfer.com/post/the-illusion-of-control-why-bare-classifier-inversion-silently-fails-in-conceptSecFeedde2026-08-25T12:02:00.147801+00:00The Illusion of Control: Why Bare Classifier Inversion Silently Fails in Concept-Bottleneck Text Generationhttps://blog.schoergendorfer.com/post/helabert-enhancing-sinhala-language-understanding-with-dual-pooling-classificatiSecFeedde2026-08-25T12:02:00.118952+00:00HelaBERT: Enhancing Sinhala Language Understanding with Dual Pooling Classification Headhttps://blog.schoergendorfer.com/post/tswap-a-multilingual-retrieval-augmented-thai-wellness-advisorSecFeedde2026-08-25T12:02:00.085151+00:00TSWAP: A Multilingual Retrieval-Augmented Thai Wellness Advisorhttps://blog.schoergendorfer.com/post/knowing-isnt-always-saying-when-do-spatial-encodings-reach-answers-in-vision-lanSecFeedde2026-08-25T12:02:00.054825+00:00Knowing Isn't Always Saying: When Do Spatial Encodings Reach Answers in Vision-Language Models?https://blog.schoergendorfer.com/post/exploring-dowker-homology-for-sentence-similaritySecFeedde2026-08-25T12:02:00.019416+00:00Exploring Dowker Homology for Sentence Similarityhttps://blog.schoergendorfer.com/post/selfusion-self-distillation-for-diffusion-language-modelsSecFeedde2026-08-25T12:01:59.983904+00:00SelFusion: Self-distillation for Diffusion Language Modelshttps://blog.schoergendorfer.com/post/better-retrieval-worse-robustnesshow-multi-hop-rag-amplifies-upstream-asr-errorsSecFeedde2026-08-25T11:46:34.104503+00:00Better Retrieval, Worse Robustness:How Multi-hop RAG Amplifies Upstream ASR Errorshttps://blog.schoergendorfer.com/post/saver-selective-auditing-of-verbal-evidence-for-error-recovery-in-vlm-change-reaSecFeedde2026-08-25T11:46:34.072685+00:00SAVER: Selective Auditing of Verbal Evidence for Error Recovery in VLM Change Reasoninghttps://blog.schoergendorfer.com/post/industrial-instruction-an-end-to-end-framework-for-building-instruction-tuning-aSecFeedde2026-08-25T11:46:34.039405+00:00Industrial-Instruction: An End-to-End Framework for Building Instruction-Tuning and Benchmark Datasets from Industrial Technical Reportshttps://blog.schoergendorfer.com/post/diag-diagnostic-iterative-alignment-and-generation-for-data-efficient-mathematicSecFeedde2026-08-25T11:46:34.002246+00:00DIAG: Diagnostic Iterative Alignment and Generation for Data-Efficient Mathematical Preference Distillationhttps://blog.schoergendorfer.com/post/spoc-sql-stage-wise-preference-optimization-for-controllable-text-to-sqlSecFeedde2026-08-25T11:46:33.971769+00:00SPOC-SQL: Stage-wise Preference Optimization for Controllable Text-to-SQLhttps://blog.schoergendorfer.com/post/delistbench-evaluating-search-enabled-llms-for-auditable-corporate-event-databasSecFeedde2026-08-25T11:46:33.941366+00:00DelistBench: Evaluating Search-Enabled LLMs for Auditable Corporate-Event Database Completionhttps://blog.schoergendorfer.com/post/beyond-factual-knowledge-benchmarking-and-learning-step-level-procedural-rule-reSecFeedde2026-08-25T11:46:33.911091+00:00Beyond Factual Knowledge: Benchmarking and Learning Step-Level Procedural Rule Reasoning in Large Language Modelshttps://blog.schoergendorfer.com/post/a-source-grounded-framework-for-constructing-and-evaluating-progressive-multimodSecFeedde2026-08-25T11:46:33.877822+00:00A Source-Grounded Framework for Constructing and Evaluating Progressive Multimodal Diagnostic Dialogues from Clinical Case Reportshttps://blog.schoergendorfer.com/post/wnw-waxing-and-waning-kv-cache-for-long-form-speech-llmsSecFeedde2026-08-25T11:46:33.843821+00:00WnW: Waxing-and-Waning KV Cache for Long-Form Speech LLMshttps://blog.schoergendorfer.com/post/from-diagnosis-to-redesign-using-quantitative-ethnography-to-improve-multi-agentSecFeedde2026-08-25T11:46:33.786588+00:00From Diagnosis to Redesign: Using Quantitative Ethnography to Improve Multi-Agent LLM Reasoninghttps://blog.schoergendorfer.com/post/kernel-token-contradiction-a-fast-and-principled-approach-for-llm-claim-uncertaiSecFeedde2026-08-25T11:46:33.753998+00:00Kernel Token Contradiction: a Fast and Principled Approach for LLM Claim Uncertainty Quantificationhttps://blog.schoergendorfer.com/post/who-pays-more-for-safety-measuring-the-disparate-cost-of-safety-alignment-acrossSecFeedde2026-08-25T11:46:33.718858+00:00Who Pays More for Safety? Measuring the Disparate Cost of Safety Alignment across Languageshttps://blog.schoergendorfer.com/post/claim-level-confidence-calibration-for-reliable-decision-making-with-large-languSecFeedde2026-08-25T11:46:33.687642+00:00Claim-Level Confidence Calibration for Reliable Decision Making with Large Language Modelshttps://blog.schoergendorfer.com/post/gta-rag-graph-trajectory-augmented-reinforcement-learning-for-multi-turn-retrievSecFeedde2026-08-25T11:46:33.657062+00:00GTA-RAG: Graph-Trajectory-Augmented Reinforcement Learning for Multi-Turn Retrieval-Augmented Reasoninghttps://blog.schoergendorfer.com/post/from-exposure-to-expectation-frequency-surprisal-and-language-across-developmentSecFeedde2026-08-25T11:46:33.625882+00:00From Exposure to Expectation: Frequency, Surprisal, and Language Across Development in Spanishhttps://blog.schoergendorfer.com/post/figurative-justice-detecting-metaphors-in-hindi-judgements-with-qualitative-asseSecFeedde2026-08-25T11:46:33.596401+00:00Figurative Justice: Detecting metaphors in Hindi judgements with qualitative assessment and transformershttps://blog.schoergendorfer.com/post/aligned-alone-misaligned-together-forecasting-adversarial-capture-in-llm-agent-pSecFeedde2026-08-25T11:46:33.558882+00:00Aligned Alone, Misaligned Together: Forecasting Adversarial Capture in LLM Agent Populationshttps://blog.schoergendorfer.com/post/don-t-box-me-in-dynamic-cultural-adaptation-and-cognitive-tracking-for-social-unSecFeedde2026-08-25T11:46:33.520809+00:00Don' t Box Me In: Dynamic Cultural Adaptation and Cognitive Tracking for Social Understandinghttps://blog.schoergendorfer.com/post/schemagui-a-schema-driven-benchmark-for-controllable-gui-generation-evaluationSecFeedde2026-08-25T11:46:33.487710+00:00SchemaGUI: A Schema-Driven Benchmark for Controllable GUI Generation Evaluationhttps://blog.schoergendorfer.com/post/can-large-language-models-hyper-threadSecFeedde2026-08-25T11:31:32.006212+00:00Can Large Language Models "Hyper-Thread"?https://blog.schoergendorfer.com/post/context-aware-cluster-decoding-semantic-anchor-driven-coherence-in-dmllmsSecFeedde2026-08-25T11:31:31.977557+00:00Context-Aware Cluster Decoding: Semantic Anchor-Driven Coherence in dMLLMshttps://blog.schoergendorfer.com/post/when-not-to-imitate-boundary-aware-skill-memory-for-reliable-tool-use-llm-agentsSecFeedde2026-08-25T11:31:31.946130+00:00When Not to Imitate: Boundary-Aware Skill Memory for Reliable Tool-Use LLM Agentshttps://blog.schoergendorfer.com/post/mechanistic-interpretability-of-chain-of-thought-reasoning-via-sequential-activaSecFeedde2026-08-25T11:31:31.911170+00:00Mechanistic Interpretability of Chain-of-Thought Reasoning via Sequential Activation Patchinghttps://blog.schoergendorfer.com/post/semantics-or-structure-auditing-text-sensitivity-in-multimodal-time-series-forecSecFeedde2026-08-25T11:31:31.853566+00:00Semantics or Structure? Auditing Text Sensitivity in Multimodal Time-Series Forecastinghttps://blog.schoergendorfer.com/post/whitewashing-hate-smearing-harmless-content-annotator-style-rebuttal-attacks-onSecFeedde2026-08-25T11:31:31.765172+00:00Whitewashing Hate, Smearing Harmless Content: Annotator-Style Rebuttal Attacks on LLM-Based Moderationhttps://blog.schoergendorfer.com/post/grounded-normative-rule-generation-with-structured-searchSecFeedde2026-08-25T11:31:31.737763+00:00Grounded Normative Rule Generation with Structured Searchhttps://blog.schoergendorfer.com/post/dual-layer-agentic-memory-with-fast-write-routing-and-slow-consolidationSecFeedde2026-08-25T11:31:31.703372+00:00Dual-Layer Agentic Memory with Fast Write Routing and Slow Consolidationhttps://blog.schoergendorfer.com/post/how-agents-represent-humans-human-directed-stereotypes-in-an-open-agent-social-nSecFeedde2026-08-25T11:31:31.671464+00:00How Agents Represent Humans: Human-Directed Stereotypes in an Open Agent Social Networkhttps://blog.schoergendorfer.com/post/the-collaboration-tax-how-much-llm-multi-agent-systems-pay-to-coordinateSecFeedde2026-08-25T11:31:31.640056+00:00The Collaboration Tax: How Much LLM Multi-Agent Systems Pay to Coordinatehttps://blog.schoergendorfer.com/post/rag-collapse-llm-responses-collapse-when-retrieved-documents-are-self-authoredSecFeedde2026-08-25T11:31:31.584742+00:00RAG Collapse: LLM Responses Collapse When Retrieved Documents Are Self-Authoredhttps://blog.schoergendorfer.com/post/spine-branch-coordination-for-multi-agent-computer-useSecFeedde2026-08-25T11:31:31.555831+00:00Spine-Branch Coordination for Multi-agent Computer Usehttps://blog.schoergendorfer.com/post/align-unify-suppress-route-a-coherentist-view-of-transformer-computationSecFeedde2026-08-25T11:31:31.526730+00:00Align, Unify, Suppress, Route: A Coherentist View of Transformer Computationhttps://blog.schoergendorfer.com/post/machine-learning-and-digital-pragmatics-which-word-category-influences-emoji-useSecFeedde2026-08-25T11:31:31.495979+00:00Machine learning and digital pragmatics: Which word category influences emoji use most?https://blog.schoergendorfer.com/post/tosca-leveraging-hierarchical-reinforcement-learning-on-temporal-and-strategic-aSecFeedde2026-08-25T11:31:31.464453+00:00ToSCA: Leveraging Hierarchical Reinforcement Learning on Temporal and Strategic Abstractions of Conversational Agentshttps://blog.schoergendorfer.com/post/modeling-claim-dependency-structure-for-patent-litigation-prediction-with-graphSecFeedde2026-08-25T11:31:31.432880+00:00Modeling Claim Dependency Structure for Patent Litigation Prediction with Graph Attention Networkshttps://blog.schoergendorfer.com/post/banglaveilguard-cross-script-safety-benchmarking-and-lightweight-guardrails-forSecFeedde2026-08-25T11:16:56.915439+00:00BanglaVeilGuard: Cross-Script Safety Benchmarking and Lightweight Guardrails for Bangla Large Language Modelshttps://blog.schoergendorfer.com/post/the-chase-is-the-curriculum-the-capture-anchors-the-credit-pursuit-evasion-selfSecFeedde2026-08-25T11:16:56.885420+00:00The Chase Is the Curriculum, the Capture Anchors the Credit: Pursuit-Evasion Self-Play for Zero-Data LLM Reasoninghttps://blog.schoergendorfer.com/post/puma-a-polish-benchmark-for-culturally-grounded-multimodal-understandingSecFeedde2026-08-25T11:16:56.856653+00:00PUMA: A Polish Benchmark for Culturally Grounded Multimodal Understandinghttps://blog.schoergendorfer.com/post/gui-primitives-diagnosing-spatial-reasoning-failures-in-vision-language-gui-grouSecFeedde2026-08-25T11:16:56.824224+00:00GUI-Primitives: Diagnosing Spatial Reasoning Failures in Vision-Language GUI Groundinghttps://blog.schoergendorfer.com/post/do-large-language-models-perform-well-on-comprehending-poetic-logic-in-modern-chSecFeedde2026-08-25T11:16:56.793724+00:00Do Large Language Models Perform Well on Comprehending Poetic Logic in Modern Chinese Poetry?https://blog.schoergendorfer.com/post/convergence-in-science-divergence-in-religion-calibrated-framing-differences-acrSecFeedde2026-08-25T11:16:56.766791+00:00Convergence in Science, Divergence in Religion: Calibrated Framing Differences Across Wikipedia's Language Editionshttps://blog.schoergendorfer.com/post/mcite-rl-towards-reliable-multimodal-rag-via-citation-enhanced-agentic-reinforceSecFeedde2026-08-25T11:16:56.739657+00:00MCite-RL: Towards Reliable Multimodal RAG via Citation-enhanced Agentic Reinforcement Learninghttps://blog.schoergendorfer.com/post/no-one-model-catches-every-harm-benchmarking-content-moderation-across-safety-scSecFeedde2026-08-25T11:16:56.713275+00:00No One Model Catches Every Harm: Benchmarking Content Moderation Across Safety Scenarioshttps://blog.schoergendorfer.com/post/fcprag-fusion-controller-parametric-retrieval-augmented-generation-for-stable-muSecFeedde2026-08-25T11:16:56.687645+00:00FCPRAG: Fusion-Controller Parametric Retrieval-Augmented Generation for Stable Multi-Passage LoRA Injectionhttps://blog.schoergendorfer.com/post/mitigating-database-leakage-in-rag-systems-with-keyword-grounded-fact-substitutiSecFeedde2026-08-25T11:16:56.627833+00:00Mitigating Database Leakage in RAG Systems with Keyword-Grounded Fact Substitutionhttps://blog.schoergendorfer.com/post/can-llms-truly-forget-revealing-unlearning-gaps-through-adversarial-evaluationSecFeedde2026-08-25T11:16:56.601072+00:00Can LLMs Truly Forget? Revealing Unlearning Gaps Through Adversarial Evaluationhttps://blog.schoergendorfer.com/post/evidence-state-reliability-under-controlled-degradation-parser-validity-divergenSecFeedde2026-08-25T11:16:56.571786+00:00Evidence-State Reliability Under Controlled Degradation: Parser-Validity Divergence in a Multi-Stage LLM Pipelinehttps://blog.schoergendorfer.com/post/towards-truly-unsupervised-evaluation-of-feature-selection-3SecFeedde2026-08-25T11:16:56.543519+00:00Towards Truly Unsupervised Evaluation of Feature Selectionhttps://blog.schoergendorfer.com/post/backdfl-a-unified-benchmark-for-backdoor-attacks-and-defenses-in-decentralized-f-2SecFeedde2026-08-25T11:16:56.512770+00:00BackDFL: A Unified Benchmark For Backdoor Attacks and Defenses In Decentralized Federated Learninghttps://blog.schoergendorfer.com/post/in-cell-learning-deployed-language-models-can-learn-new-knowledge-without-changiSecFeedde2026-08-25T11:16:56.485113+00:00In-Cell Learning: Deployed Language Models Can Learn New Knowledge Without Changing a Single Stored Bithttps://blog.schoergendorfer.com/post/atom-learning-model-alm-how-a-real-classroom-got-tokenised-2SecFeedde2026-08-25T11:16:56.452644+00:00Atom Learning Model (ALM): how a real classroom got tokenisedhttps://blog.schoergendorfer.com/post/at-vit-area-targeted-multi-view-vision-transformer-with-cross-attention-and-mult-2SecFeedde2026-08-25T11:16:56.424755+00:00AT-ViT: Area-Targeted Multi-View Vision Transformer with Cross-Attention and Multi-Scale Patching for Plant Trait Recognition in Herbarium Imageshttps://blog.schoergendorfer.com/post/task-coevolve-efficient-harness-optimization-via-adaptive-validation-task-select-2SecFeedde2026-08-25T11:16:56.393034+00:00Task-CoEvolve: Efficient Harness Optimization via Adaptive Validation Task Selectionhttps://blog.schoergendorfer.com/post/milegpo-milestone-inference-with-local-evidence-for-graph-based-policy-optimizat-2SecFeedde2026-08-25T11:16:56.362772+00:00MileGPO: Milestone Inference with Local Evidence for Graph-Based Policy Optimization of Long-Horizon LLM Agentshttps://blog.schoergendorfer.com/post/question-guided-evidence-acquisition-for-multimodal-visual-question-answering-2SecFeedde2026-08-25T11:01:49.924124+00:00Question-Guided Evidence Acquisition for Multimodal Visual Question Answeringhttps://blog.schoergendorfer.com/post/learning-to-beat-phenotype-guided-latent-flow-with-regional-motion-priors-for-bi-2SecFeedde2026-08-25T11:01:49.892078+00:00Learning to Beat: Phenotype-Guided Latent Flow with Regional Motion Priors for Biventricular Motion Synthesishttps://blog.schoergendorfer.com/post/vgi-bench-probing-visual-intelligence-in-video-generation-models-2SecFeedde2026-08-25T11:01:49.856030+00:00VGI-Bench: Probing Visual Intelligence in Video Generation Modelshttps://blog.schoergendorfer.com/post/active-spiking-perception-the-membrane-potential-as-a-belief-state-for-anytime-3-2SecFeedde2026-08-25T11:01:49.822375+00:00Active Spiking Perception: The Membrane Potential as a Belief State for Anytime 3D Point Cloud Recognitionhttps://blog.schoergendorfer.com/post/time-series-retrieval-for-grounding-multimodal-language-models-in-remaining-usef-2SecFeedde2026-08-25T11:01:49.789215+00:00Time-Series Retrieval for Grounding Multimodal Language Models in Remaining Useful Life Predictionhttps://blog.schoergendorfer.com/post/spade-self-play-in-adaptive-synthetic-executable-environments-2SecFeedde2026-08-25T11:01:49.753706+00:00SPADE: Self-Play in Adaptive Synthetic Executable Environmentshttps://blog.schoergendorfer.com/post/interpretable-ai-predicts-a-2026-summer-dry-anomaly-in-central-china-2SecFeedde2026-08-25T11:01:49.722545+00:00Interpretable AI predicts a 2026 summer dry anomaly in central Chinahttps://blog.schoergendorfer.com/post/alphaclifford-efficient-clifford-synthesis-and-transpilation-with-model-based-rl-2SecFeedde2026-08-25T11:01:49.684297+00:00AlphaClifford: Efficient Clifford Synthesis and Transpilation with Model-based RLhttps://blog.schoergendorfer.com/post/aslema-at-nadi-2026-data-augmentation-for-intent-recognition-and-slot-fillingSecFeedde2026-08-25T11:01:49.650053+00:00Aslema at NADI 2026: Data Augmentation for Intent Recognition and Slot Fillinghttps://blog.schoergendorfer.com/post/formal-verification-of-romanovs-triplet-logic-a-verified-filter-for-sliding-wind-2SecFeedde2026-08-25T11:01:49.616001+00:00Formal Verification of Romanov's Triplet Logic: A Verified Filter for Sliding-window 3-CNF with Application to Structured Formulashttps://blog.schoergendorfer.com/post/gigabrain-wbc-05-a-behavior-world-model-for-robust-whole-body-control-with-envir-2SecFeedde2026-08-25T11:01:49.583659+00:00GigaBrain-WBC-0.5: A Behavior World Model for Robust Whole-Body Control with Environment Interactionhttps://blog.schoergendorfer.com/post/coal-rag-a-complexity-aware-legal-retrieval-augmented-generation-method-2SecFeedde2026-08-25T11:01:49.552879+00:00CoAL-RAG: A Complexity-Aware Legal Retrieval-Augmented Generation Methodhttps://blog.schoergendorfer.com/post/pertmind-eliciting-emergent-biological-reasoning-in-llm-via-reinforcement-learni-2SecFeedde2026-08-25T11:01:49.515721+00:00PertMind: Eliciting Emergent Biological Reasoning in LLM via Reinforcement Learning on Cellular Perturbation Datahttps://blog.schoergendorfer.com/post/learning-to-unlearn-machine-unlearning-via-learning-the-unlearning-behaviors-2SecFeedde2026-08-25T11:01:49.477901+00:00Learning to Unlearn: Machine Unlearning via Learning the Unlearning Behaviorshttps://blog.schoergendorfer.com/post/granikv-asymmetric-granularity-kv-cache-paging-for-multi-agent-systems-with-long-2SecFeedde2026-08-25T11:01:49.433171+00:00GraniKV: Asymmetric Granularity KV-Cache Paging for Multi-Agent Systems with Long Shared Prefixhttps://blog.schoergendorfer.com/post/emergent-misaligned-communication-in-long-horizon-multi-agent-llm-commerce-3SecFeedde2026-08-25T11:01:49.397448+00:00Emergent Misaligned Communication in Long-Horizon Multi-Agent LLM Commercehttps://blog.schoergendorfer.com/post/patientact-theory-grounded-mental-health-client-simulation-2SecFeedde2026-08-25T11:01:49.362106+00:00PatientAct: Theory-Grounded Mental Health Client Simulationhttps://blog.schoergendorfer.com/post/withholding-the-completing-chunk-exact-release-boundary-equivalence-for-productiSecFeedde2026-08-25T11:01:49.324091+00:00Withholding the Completing Chunk: Exact Release-Boundary Equivalence for Production Streaming Guardrailshttps://blog.schoergendorfer.com/post/timeroute-time-aware-modality-routing-and-diffusion-for-multi-modal-recommendati-2SecFeedde2026-08-25T11:01:49.290680+00:00TimeRoute: Time-Aware Modality Routing and Diffusion for Multi-Modal Recommendationhttps://blog.schoergendorfer.com/post/population-scalable-multi-agent-world-modeling-3SecFeedde2026-08-25T11:01:49.257951+00:00Population-Scalable Multi-Agent World Modelinghttps://blog.schoergendorfer.com/post/hidden-language-consistency-phenomena-in-reasoning-llms-2SecFeedde2026-08-25T10:46:36.341222+00:00Hidden Language Consistency Phenomena in Reasoning LLMshttps://blog.schoergendorfer.com/post/answer-first-reason-later-when-commitment-order-costs-accuracy-in-diffusion-langSecFeedde2026-08-25T10:46:36.309609+00:00Answer First, Reason Later: When Commitment Order Costs Accuracy in Diffusion Language Modelshttps://blog.schoergendorfer.com/post/breadcrumbing-search-agents-2SecFeedde2026-08-25T10:46:36.276157+00:00Breadcrumbing Search Agentshttps://blog.schoergendorfer.com/post/svi-dag-a-structured-variational-inference-approach-to-bayesian-causal-discovery-2SecFeedde2026-08-25T10:46:36.239222+00:00SVI-DAG: A Structured Variational Inference Approach to Bayesian Causal Discoveryhttps://blog.schoergendorfer.com/post/ai-alignment-and-fiduciary-obligation-2SecFeedde2026-08-25T10:46:36.205858+00:00AI Alignment and Fiduciary Obligationhttps://blog.schoergendorfer.com/post/your-agentic-llms-secretly-encode-indirect-prompt-injection-exposure-in-hidden-sSecFeedde2026-08-25T10:46:36.178831+00:00Your Agentic LLMs Secretly Encode Indirect Prompt-Injection Exposure in Hidden Stateshttps://blog.schoergendorfer.com/post/measuring-in-context-algorithmic-reasoning-in-language-models-against-an-exact-b-2SecFeedde2026-08-25T10:46:36.148413+00:00Measuring in-context algorithmic reasoning in language models against an exact Bayes-optimal referencehttps://blog.schoergendorfer.com/post/divisive-normalization-shapes-low-rank-slow-manifolds-for-continuous-working-memSecFeedde2026-08-25T10:46:36.114293+00:00Divisive Normalization Shapes Low-Rank Slow Manifolds for Continuous Working Memoryhttps://blog.schoergendorfer.com/post/callscreenbench-benchmarking-small-language-models-as-phone-secretariesSecFeedde2026-08-25T10:46:36.077705+00:00CallScreenBench: Benchmarking Small Language Models as Phone Secretarieshttps://blog.schoergendorfer.com/post/latency-tolerant-cloud-edge-collaborative-vision-language-action-models-via-emerSecFeedde2026-08-25T10:46:36.040358+00:00Latency-Tolerant Cloud-Edge Collaborative Vision-Language-Action Models via Emergent Representational Specializationhttps://blog.schoergendorfer.com/post/what-transfers-from-text-to-vision-capability-scaling-laws-and-transfer-dynamics-2SecFeedde2026-08-25T10:46:36.011837+00:00What Transfers from Text to Vision? Capability Scaling Laws and Transfer Dynamics for VLMshttps://blog.schoergendorfer.com/post/idp-autoopt-agent-driven-optimization-of-document-processing-pipeline-configurat-2SecFeedde2026-08-25T10:46:35.974996+00:00IDP AutoOpt: Agent-Driven Optimization of Document Processing Pipeline Configurationshttps://blog.schoergendorfer.com/post/a2tta-anchored-and-agile-test-time-adaptation-for-evolving-traffic-sensor-networ-3SecFeedde2026-08-25T10:46:35.942233+00:00A2TTA: Anchored-and-Agile Test-Time Adaptation for Evolving Traffic Sensor Networkshttps://blog.schoergendorfer.com/post/a-formal-kinetic-theory-for-zeroth-order-newton-dynamicsstein-corrected-hessian-2SecFeedde2026-08-25T10:46:35.909551+00:00A Formal Kinetic Theory for Zeroth-Order Newton Dynamics:Stein-Corrected Hessian Estimation and Curvature--Variance Trade-offshttps://blog.schoergendorfer.com/post/causalsmith-a-formally-grounded-self-improving-agentic-framework-for-automated-rSecFeedde2026-08-25T10:46:35.878201+00:00CausalSmith: A Formally Grounded, Self-Improving Agentic Framework for Automated Research in Causal Inferencehttps://blog.schoergendorfer.com/post/candor-chance-calibrated-discordance-in-frozen-foundation-encoders-2SecFeedde2026-08-25T10:46:35.847210+00:00CANDOR: Chance-Calibrated Discordance in Frozen Foundation Encodershttps://blog.schoergendorfer.com/post/pathagentbench-benchmarking-evidence-seeking-vision-language-models-on-whole-sli-2SecFeedde2026-08-25T10:46:35.811917+00:00PathAgentBench: Benchmarking Evidence-Seeking Vision-Language Models on Whole-Slide Pathology Imagehttps://blog.schoergendorfer.com/post/adversarial-robustness-of-phishing-email-detection-a-comparative-study-of-tf-idf-2SecFeedde2026-08-25T10:46:35.779942+00:00Adversarial Robustness of Phishing Email Detection: A Comparative Study of TF-IDF + Logistic Regression and Fine-Tuned DistilBERThttps://blog.schoergendorfer.com/post/falsification-based-verification-of-llm-generated-optimization-models-sound-test-2SecFeedde2026-08-25T10:46:35.745485+00:00Falsification-Based Verification of LLM-Generated Optimization Models: Sound Test Batteries and Their Detection Limitshttps://blog.schoergendorfer.com/post/silent-alarm-a-j-space-protocol-for-comparing-danger-recognition-across-models-a-2SecFeedde2026-08-25T10:31:59.514646+00:00Silent Alarm: A J-Space Protocol for Comparing Danger Recognition Across Models and Quantization Levelshttps://blog.schoergendorfer.com/post/grc-probnet-uncertainty-aware-feature-extraction-for-cardiovascular-disease-clas-2SecFeedde2026-08-25T10:31:59.452291+00:00GRC-ProbNet: Uncertainty-aware Feature Extraction for Cardiovascular Disease Classificationhttps://blog.schoergendorfer.com/post/diallm-an-investigation-into-the-robustness-generation-gap-in-english-dialect-ad-2SecFeedde2026-08-25T10:31:59.425262+00:00DiaLLM: An Investigation into the Robustness-Generation Gap in English Dialect Adaptationhttps://blog.schoergendorfer.com/post/post-generation-curation-of-synthetic-images-via-homogeneous-heterogeneous-split-2SecFeedde2026-08-25T10:31:59.399742+00:00Post-Generation Curation of Synthetic Images via Homogeneous-Heterogeneous Splittinghttps://blog.schoergendorfer.com/post/guided-action-flow-q-guided-inference-for-flow-matching-vision-language-action-p-2SecFeedde2026-08-25T10:31:59.368368+00:00Guided Action Flow: Q-Guided Inference for Flow-Matching Vision-Language-Action Policieshttps://blog.schoergendorfer.com/post/compositional-dynamics-in-learning-and-mechanicsSecFeedde2026-08-25T10:31:59.342068+00:00Compositional Dynamics in Learning and Mechanicshttps://blog.schoergendorfer.com/post/group-equivariant-poincare-convolutional-networksSecFeedde2026-08-25T10:31:59.316767+00:00Group-Equivariant Poincar\'e Convolutional Networkshttps://blog.schoergendorfer.com/post/what-llms-explain-is-not-what-they-believe-evaluating-explanation-sufficiency-unSecFeedde2026-08-25T10:31:59.291606+00:00What LLMs explain is not what they believe: Evaluating explanation sufficiency under models' own input beliefshttps://blog.schoergendorfer.com/post/event-conditioned-diagnostics-of-kinematic-contact-and-object-permanence-structuSecFeedde2026-08-25T10:31:59.260827+00:00Event-Conditioned Diagnostics of Kinematic, Contact, and Object-Permanence Structure in Passive Object-State World Modelshttps://blog.schoergendorfer.com/post/rarm-confidence-gated-progress-reward-modeling-for-rl-in-manipulationSecFeedde2026-08-25T10:31:59.232757+00:00RARM: Confidence-Gated Progress Reward Modeling for RL in Manipulationhttps://blog.schoergendorfer.com/post/one-polluted-page-is-enough-evaluating-web-content-pollution-in-llm-recommendersSecFeedde2026-08-25T10:31:59.143556+00:00One Polluted Page Is Enough: Evaluating Web Content Pollution in LLM Recommendershttps://blog.schoergendorfer.com/post/chain-of-operators-an-inference-time-harness-for-in-context-operator-learningSecFeedde2026-08-25T10:31:59.110909+00:00Chain of Operators: An Inference-Time Harness for In-Context Operator Learninghttps://blog.schoergendorfer.com/post/trace-a-unified-rollout-budget-allocation-framework-for-efficient-agentic-reinfoSecFeedde2026-08-25T10:31:59.079236+00:00TRACE: A Unified Rollout Budget Allocation Framework for Efficient Agentic Reinforcement Learninghttps://blog.schoergendorfer.com/post/toward-secure-llm-agents-threat-surfaces-attacks-defenses-and-evaluationSecFeedde2026-08-25T10:31:59.048339+00:00Toward Secure LLM Agents: Threat Surfaces, Attacks, Defenses, and Evaluationhttps://blog.schoergendorfer.com/post/socraticpo-policy-optimization-via-interactive-guidanceSecFeedde2026-08-25T10:31:59.014341+00:00SocraticPO: Policy Optimization via Interactive Guidancehttps://blog.schoergendorfer.com/post/e2llm-towards-efficient-llm-serving-in-heterogeneous-edgefog-environments-2SecFeedde2026-08-25T10:31:58.953330+00:00E2LLM: Towards Efficient LLM Serving in Heterogeneous Edge/Fog Environmentshttps://blog.schoergendorfer.com/post/soft-nbce-entropy-weighted-chunk-fusion-for-long-contextSecFeedde2026-08-25T10:16:40.911624+00:00Soft-NBCE: Entropy-Weighted Chunk Fusion for Long-Contexthttps://blog.schoergendorfer.com/post/drift-joint-channel-estimation-and-prediction-towards-pilotless-6g-non-terrestriSecFeedde2026-08-25T10:16:40.882243+00:00DRIFT: Joint Channel Estimation and Prediction Towards Pilotless 6G Non-Terrestrial Networkshttps://blog.schoergendorfer.com/post/birdnet-mining-and-encoding-boolean-implication-knowledge-graphs-as-interpretablSecFeedde2026-08-25T10:16:40.848171+00:00BIRDNet: Mining and Encoding Boolean Implication Knowledge Graphs as Interpretable Deep Neural Networkshttps://blog.schoergendorfer.com/post/how-should-llms-consume-high-quality-data-optimal-data-scheduling-via-quality-awSecFeedde2026-08-25T10:16:40.811222+00:00How Should LLMs Consume High-Quality Data? Optimal Data Scheduling via Quality-Aware Functional Scaling Lawshttps://blog.schoergendorfer.com/post/edge-ai-driven-learning-to-rank-for-decentralized-task-allocation-in-circular-smSecFeedde2026-08-25T10:16:40.780828+00:00Edge-AI-Driven Learning-to-Rank for Decentralized Task Allocation in Circular Smart Manufacturinghttps://blog.schoergendorfer.com/post/revisiting-reinforcement-learning-with-verifiable-rewards-from-a-contrastive-perSecFeedde2026-08-25T10:16:40.752450+00:00Revisiting Reinforcement Learning with Verifiable Rewards from a Contrastive Perspectivehttps://blog.schoergendorfer.com/post/deeprefine-agentic-knowledge-refinement-via-reinforcement-learningSecFeedde2026-08-25T10:16:40.724313+00:00DeepRefine: Agentic Knowledge Refinement via Reinforcement Learninghttps://blog.schoergendorfer.com/post/dprm-a-plug-in-doob-h-transform-induced-token-ordering-module-for-discrete-diffuSecFeedde2026-08-25T10:16:40.694136+00:00DPRM: A Plug-in Doob h transform-induced Token-Ordering Module for Discrete Diffusion Modelshttps://blog.schoergendorfer.com/post/onote-hypergraph-grounded-omnimodal-reasoning-for-computational-music-scienceSecFeedde2026-08-25T10:16:40.666186+00:00ONOTE: Hypergraph-Grounded Omnimodal Reasoning for Computational Music Sciencehttps://blog.schoergendorfer.com/post/dialtom-a-theory-of-mind-benchmark-for-forecasting-state-driven-dialogue-trajectSecFeedde2026-08-25T10:16:40.633896+00:00DialToM: A Theory of Mind Benchmark for Forecasting State-Driven Dialogue Trajectorieshttps://blog.schoergendorfer.com/post/conscientia-can-llm-agents-learn-to-strategize-emergent-deception-and-trust-in-aSecFeedde2026-08-25T10:16:40.575603+00:00CONSCIENTIA: Can LLM Agents Learn to Strategize? Emergent Deception and Trust in a Multi-Agent NYC Simulationhttps://blog.schoergendorfer.com/post/large-language-models-generate-harmful-responses-using-a-distinct-mechanism-shar-2SecFeedde2026-08-25T10:16:40.543326+00:00Large Language Models Generate Harmful Responses Using a Distinct Mechanism, Shared Across Harm Typeshttps://blog.schoergendorfer.com/post/beyond-rgb-benchmarking-and-enhancing-mllms-for-hyperspectral-image-understandinSecFeedde2026-08-25T10:16:40.513059+00:00Beyond RGB: Benchmarking and Enhancing MLLMs for Hyperspectral Image Understanding via Training-Free Reasoning Frameworkhttps://blog.schoergendorfer.com/post/verbalizing-llms-assumptions-to-explain-and-control-sycophancySecFeedde2026-08-25T10:16:40.424089+00:00Verbalizing LLMs' assumptions to explain and control sycophancyhttps://blog.schoergendorfer.com/post/robust-multi-agent-reinforcement-learning-for-small-uas-separation-assurance-undSecFeedde2026-08-25T10:16:40.372123+00:00Robust Multi-Agent Reinforcement Learning for Small UAS Separation Assurance under GPS Degradation and Spoofinghttps://blog.schoergendorfer.com/post/when-should-a-robot-think-resource-aware-reasoning-via-reinforcement-learning-foSecFeedde2026-08-25T10:16:40.343362+00:00When Should a Robot Think? Resource-Aware Reasoning via Reinforcement Learning for Embodied Robotic Decision-Makinghttps://blog.schoergendorfer.com/post/visioncoach-reinforcing-grounded-video-reasoning-via-visual-perception-promptingSecFeedde2026-08-25T10:01:41.447447+00:00VisionCoach: Reinforcing Grounded Video Reasoning via Visual-Perception Promptinghttps://blog.schoergendorfer.com/post/semantic-substrate-dynamics-theory-an-operator-theoretic-framework-for-geometricSecFeedde2026-08-25T10:01:41.417243+00:00Semantic Substrate Dynamics Theory: An Operator-Theoretic Framework for Geometric Semantic Drifthttps://blog.schoergendorfer.com/post/vibe-coding-on-trial-operating-characteristics-of-unanimous-llm-juriesSecFeedde2026-08-25T10:01:41.383615+00:00Vibe Coding on Trial: Operating Characteristics of Unanimous LLM Jurieshttps://blog.schoergendorfer.com/post/learning-with-boolean-threshold-functionsSecFeedde2026-08-25T10:01:41.349321+00:00Learning with Boolean threshold functionshttps://blog.schoergendorfer.com/post/continual-uncertainty-learning-for-robust-control-of-nonlinear-systems-with-multSecFeedde2026-08-25T10:01:41.315247+00:00Continual Uncertainty Learning for Robust Control of Nonlinear Systems with Multiple Heterogeneous Uncertaintieshttps://blog.schoergendorfer.com/post/which-algorithms-can-graph-neural-networks-learnSecFeedde2026-08-25T10:01:41.247572+00:00Which Algorithms Can Graph Neural Networks Learn?https://blog.schoergendorfer.com/post/ischeduler-reinforcement-learning-driven-continual-optimization-for-large-scaleSecFeedde2026-08-25T10:01:41.212087+00:00iScheduler: Reinforcement Learning-Driven Continual Optimization for Large-Scale Resource Investment Problemshttps://blog.schoergendorfer.com/post/backdoor-sentinel-detecting-and-detoxifying-backdoors-in-diffusion-models-via-teSecFeedde2026-08-25T10:01:41.098172+00:00Backdoor Sentinel: Detecting and Detoxifying Backdoors in Diffusion Models via Temporal Noise Consistencyhttps://blog.schoergendorfer.com/post/dynamic-cogeneration-of-bug-reproduction-test-in-agentic-program-repairSecFeedde2026-08-25T10:01:41.008195+00:00Dynamic Cogeneration of Bug Reproduction Test in Agentic Program Repairhttps://blog.schoergendorfer.com/post/peafowl-perception-enhanced-multi-view-vision-language-action-for-bimanual-manipSecFeedde2026-08-25T10:01:40.973789+00:00PEAfowl: Perception-Enhanced Multi-View Vision-Language-Action for Bimanual Manipulationhttps://blog.schoergendorfer.com/post/benchmarking-document-parsers-on-mathematical-formula-extraction-from-pdfsSecFeedde2026-08-25T10:01:40.889671+00:00Benchmarking Document Parsers on Mathematical Formula Extraction from PDFshttps://blog.schoergendorfer.com/post/diagnosing-capability-preservation-and-task-sensitivity-in-memory-augmented-docuSecFeedde2026-08-25T10:01:40.859505+00:00Diagnosing Capability Preservation and Task Sensitivity in Memory Augmented Document Classifiershttps://blog.schoergendorfer.com/post/mpcformer-a-physics-informed-data-driven-approach-for-explainable-socially-awareSecFeedde2026-08-25T09:46:51.664527+00:00MPCFormer: A physics-informed data-driven approach for explainable socially-aware autonomous drivinghttps://blog.schoergendorfer.com/post/multi-context-fusion-transformer-for-pedestrian-crossing-intention-prediction-inSecFeedde2026-08-25T09:46:51.633881+00:00Multi-Context Fusion Transformer for Pedestrian Crossing Intention Prediction in Urban Environmentshttps://blog.schoergendorfer.com/post/moclip-a-foundation-model-for-large-scale-nanophotonic-inverse-designSecFeedde2026-08-25T09:46:51.607086+00:00MOCLIP: A Foundation Model for Large-Scale Nanophotonic Inverse Designhttps://blog.schoergendorfer.com/post/hifinet-hierarchical-fault-identification-in-wireless-sensor-networks-via-edge-bSecFeedde2026-08-25T09:46:51.581643+00:00HiFiNet: Hierarchical Fault Identification in Wireless Sensor Networks via Edge-Based Classification and Graph Aggregationhttps://blog.schoergendorfer.com/post/radial-compensation-the-inverse-base-distribution-problem-for-chart-based-generaSecFeedde2026-08-25T09:46:51.549249+00:00Radial Compensation: The Inverse Base-Distribution Problem for Chart-Based Generative Models on Riemannian Manifoldshttps://blog.schoergendorfer.com/post/one-request-multiple-experts-llm-orchestrates-domain-specific-models-via-adaptivSecFeedde2026-08-25T09:46:51.514628+00:00One Request, Multiple Experts: LLM Orchestrates Domain Specific Models via Adaptive Task Routinghttps://blog.schoergendorfer.com/post/a-new-type-of-adversarial-examplesSecFeedde2026-08-25T09:46:51.483687+00:00A New Type of Adversarial Exampleshttps://blog.schoergendorfer.com/post/mitigating-sample-level-imbalance-via-probabilistic-separation-for-adaptive-multSecFeedde2026-08-25T09:46:51.454600+00:00Mitigating Sample-Level Imbalance via Probabilistic Separation for Adaptive Multimodal Fusionhttps://blog.schoergendorfer.com/post/llm-specific-utility-for-retrieval-augmented-generationSecFeedde2026-08-25T09:46:51.426824+00:00LLM-Specific Utility for Retrieval-Augmented Generationhttps://blog.schoergendorfer.com/post/graphmed-lt-patient-specific-graph-memory-with-latent-clinical-thought-refinemenSecFeedde2026-08-25T09:46:51.399410+00:00GraphMed-LT: Patient-Specific Graph Memory with Latent Clinical Thought Refinement for Multi-Turn Medical Conversationshttps://blog.schoergendorfer.com/post/slogic-subgraph-informed-logical-rule-learning-for-knowledge-graph-completionSecFeedde2026-08-25T09:46:51.371008+00:00SLogic: Subgraph-Informed Logical Rule Learning for Knowledge Graph Completionhttps://blog.schoergendorfer.com/post/rstgcn-railway-centric-spatio-temporal-graph-convolutional-network-for-train-delSecFeedde2026-08-25T09:46:51.340362+00:00RSTGCN: Railway-centric Spatio-Temporal Graph Convolutional Network for Train Delay Predictionhttps://blog.schoergendorfer.com/post/hivis-hiding-visual-tokens-from-the-drafter-for-speculative-decoding-in-vision-lSecFeedde2026-08-25T09:46:51.309755+00:00HiViS: Hiding Visual Tokens from the Drafter for Speculative Decoding in Vision-Language Modelshttps://blog.schoergendorfer.com/post/extrinsplat-decoupling-geometry-and-semantics-for-open-vocabulary-understandingSecFeedde2026-08-25T09:46:51.284727+00:00ExtrinSplat: Decoupling Geometry and Semantics for Open-Vocabulary Understanding in 3D Gaussian Splattinghttps://blog.schoergendorfer.com/post/an-information-flow-perspective-on-explainability-requirements-specification-andSecFeedde2026-08-25T09:46:51.256706+00:00An Information-Flow Perspective on Explainability Requirements: Specification and Verificationhttps://blog.schoergendorfer.com/post/entity-representation-learning-through-onsite-offsite-graph-for-pinterest-adsSecFeedde2026-08-25T09:46:51.229559+00:00Entity Representation Learning Through Onsite-Offsite Graph for Pinterest Adshttps://blog.schoergendorfer.com/post/from-isolation-to-alignment-unified-lora-for-efficient-multi-task-learningSecFeedde2026-08-25T09:46:51.200752+00:00From Isolation to Alignment: Unified LoRA for Efficient Multi-Task Learninghttps://blog.schoergendorfer.com/post/televal-a-benchmark-designed-for-spoken-language-models-in-chinese-interactive-sSecFeedde2026-08-25T09:46:51.172076+00:00TELEVAL: A Benchmark Designed for Spoken Language Models in Chinese Interactive Scenarioshttps://blog.schoergendorfer.com/post/text-adbench-text-anomaly-detection-benchmark-based-on-llm-embeddingsSecFeedde2026-08-25T09:46:51.116637+00:00Text-ADBench: Text Anomaly Detection Benchmark Based on LLM Embeddingshttps://blog.schoergendorfer.com/post/mission-aligned-learning-informed-control-of-autonomous-systems-formulation-andSecFeedde2026-08-25T09:31:50.108831+00:00Mission-Aligned Learning-Informed Control of Autonomous Systems: Formulation and Foundationshttps://blog.schoergendorfer.com/post/a-modular-multitask-reasoning-framework-integrating-spatio-temporal-models-and-lSecFeedde2026-08-25T09:31:50.078202+00:00A Modular Multitask Reasoning Framework Integrating Spatio-temporal Models and LLMshttps://blog.schoergendorfer.com/post/from-recognition-to-reasoning-advancing-multimodal-harmful-meme-detection-via-chSecFeedde2026-08-25T09:31:50.045938+00:00From Recognition to Reasoning: Advancing Multimodal Harmful Meme Detection via Chain-of-Thought Alignmenthttps://blog.schoergendorfer.com/post/seismic-acoustic-impedance-inversion-framework-based-on-conditional-latent-generSecFeedde2026-08-25T09:31:50.015859+00:00Seismic Acoustic Impedance Inversion Framework Based on Conditional Latent Generative Diffusion Modelhttps://blog.schoergendorfer.com/post/time-series-forecasting-via-reasoning-a-slow-thinking-approach-with-reinforcemenSecFeedde2026-08-25T09:31:49.984197+00:00Time Series Forecasting via Reasoning: A Slow-Thinking Approach with Reinforcement Fine-Tuned LLMshttps://blog.schoergendorfer.com/post/effects-of-theory-of-mind-and-prosocial-beliefs-on-steering-human-aligned-behaviSecFeedde2026-08-25T09:31:49.956431+00:00Effects of Theory of Mind and Prosocial Beliefs on Steering Human-Aligned Behaviors of LLMs in Ultimatum Gameshttps://blog.schoergendorfer.com/post/colora-efficient-fine-tuning-for-convolutional-models-with-a-study-case-on-optic-2SecFeedde2026-08-25T09:31:49.898793+00:00COLORA: Efficient Fine-Tuning for Convolutional Models with a Study Case on Optical Coherence Tomography Image Classificationhttps://blog.schoergendorfer.com/post/unleashing-the-power-of-llms-in-dense-retrieval-with-query-likelihood-modelingSecFeedde2026-08-25T09:31:49.843154+00:00Unleashing the Power of LLMs in Dense Retrieval with Query Likelihood Modelinghttps://blog.schoergendorfer.com/post/ai-university-an-llm-powered-learning-assistant-for-engineering-a-finite-elementSecFeedde2026-08-25T09:31:49.817572+00:00AI University: An LLM-Powered Learning Assistant for Engineering---A Finite Element Method Case Studyhttps://blog.schoergendorfer.com/post/deep-contrastive-unlearning-for-language-modelsSecFeedde2026-08-25T09:31:49.787361+00:00Deep Contrastive Unlearning for Language Modelshttps://blog.schoergendorfer.com/post/sas-segment-anything-small-for-ultrasound-a-non-generative-data-augmentation-tecSecFeedde2026-08-25T09:31:49.757769+00:00SAS: Segment Anything Small for Ultrasound -- A Non-Generative Data Augmentation Technique for Robust Deep Learning in Ultrasound Imaginghttps://blog.schoergendorfer.com/post/srmt-shared-memory-for-multi-agent-lifelong-pathfindingSecFeedde2026-08-25T09:31:49.675367+00:00SRMT: Shared Memory for Multi-agent Lifelong Pathfindinghttps://blog.schoergendorfer.com/post/nest-neighborhood-aware-semantic-alignment-and-temporal-modulation-for-llm-basedSecFeedde2026-08-25T09:31:49.647294+00:00NeST: Neighborhood-aware semantic alignment and temporal modulation for LLM based time series forecastinghttps://blog.schoergendorfer.com/post/revisiting-multi-permutation-equivariance-through-the-lens-of-irreducible-repres-2SecFeedde2026-08-25T09:31:49.621127+00:00Revisiting Multi-Permutation Equivariance through the Lens of Irreducible Representationshttps://blog.schoergendorfer.com/post/lsem2vec-a-simple-yet-effective-two-stage-approach-for-source-code-embedding-2SecFeedde2026-08-25T09:31:49.590703+00:00LSem2Vec: A Simple yet Effective Two-Stage Approach for Source Code Embeddinghttps://blog.schoergendorfer.com/post/image-conditional-diffusion-transformer-for-underwater-image-enhancementSecFeedde2026-08-25T09:17:09.811161+00:00Image-Conditional Diffusion Transformer for Underwater Image Enhancementhttps://blog.schoergendorfer.com/post/evaluating-the-efficacy-of-llms-to-emulate-realistic-human-personalitiesSecFeedde2026-08-25T09:17:09.783855+00:00Evaluating the Efficacy of LLMs to Emulate Realistic Human Personalitieshttps://blog.schoergendorfer.com/post/towards-a-resource-for-multilingual-lexicons-an-mt-assisted-and-human-in-the-looSecFeedde2026-08-25T09:17:09.753874+00:00Towards a resource for multilingual lexicons: an MT assisted and human-in-the-loop multilingual parallel corpus with multi-word expression annotationhttps://blog.schoergendorfer.com/post/why-we-need-an-ai-resilient-society-profiling-large-language-modelsSecFeedde2026-08-25T09:17:09.695617+00:00Why we need an AI-resilient society- Profiling Large Language Modelshttps://blog.schoergendorfer.com/post/foretime-vla-causal-future-token-distillation-from-a-world-action-model-for-conv-2SecFeedde2026-08-25T09:17:09.643502+00:00ForeTime-VLA: Causal Future-Token Distillation from a World Action Model for Conveyor-Belt Manipulationhttps://blog.schoergendorfer.com/post/pandoras-ai-model-routing-box-efficient-allocation-with-costly-value-estimation-2SecFeedde2026-08-25T09:17:09.616155+00:00Pandora's AI Model Routing Box: Efficient Allocation with Costly Value Estimationhttps://blog.schoergendorfer.com/post/what-you-cant-see-is-what-you-learn-restricted-evidence-visibility-favors-compos-2SecFeedde2026-08-25T09:17:09.589483+00:00What You Can't See Is What You Learn: Restricted Evidence Visibility Favors Compositional Generalization in Shared-Genome Language-Model Societieshttps://blog.schoergendorfer.com/post/autoresearch-insight-in-hallucination-out-2SecFeedde2026-08-25T09:17:09.529230+00:00AutoResearch: Insight In, Hallucination Outhttps://blog.schoergendorfer.com/post/skilleffect-checked-lowering-for-memory-bounded-agent-tools-2SecFeedde2026-08-25T09:17:09.500237+00:00SkillEffect: Checked Lowering for Memory-Bounded Agent Toolshttps://blog.schoergendorfer.com/post/harecap-habitual-action-grounding-for-recursive-large-language-model-agents-2SecFeedde2026-08-25T09:17:09.469512+00:00HaReCAP: Habitual-action Grounding for Recursive Large Language Model Agentshttps://blog.schoergendorfer.com/post/when-entropy-is-not-enough-reclaiming-lost-semantics-in-llm-output-length-predic-2SecFeedde2026-08-25T09:17:09.438665+00:00When Entropy Is Not Enough: Reclaiming Lost Semantics in LLM Output Length Predictionhttps://blog.schoergendorfer.com/post/framework-for-grounding-healthcare-llms-in-a-causal-knowledge-graph-a-cardiovascSecFeedde2026-08-25T09:17:09.345395+00:00Framework for Grounding Healthcare LLMs in a Causal Knowledge Graph: A Cardiovascular Example Pilothttps://blog.schoergendorfer.com/post/scienceflow-a-long-horizon-agent-for-ml-research-scientific-discovery-and-beyond-2SecFeedde2026-08-25T09:17:09.314176+00:00ScienceFlow: A long-horizon agent for ML research, scientific discovery and beyondhttps://blog.schoergendorfer.com/post/jagged-judges-epistemic-stability-under-perturbation-pressure-and-persistenceSecFeedde2026-08-25T09:17:09.276320+00:00Jagged Judges: Epistemic Stability Under Perturbation, Pressure, and Persistencehttps://blog.schoergendorfer.com/post/infrabench-evaluating-infrastructure-agents-across-layers-lifecycle-and-risk-2SecFeedde2026-08-25T09:17:09.246599+00:00InfraBench: Evaluating Infrastructure Agents Across Layers, Lifecycle, and Riskhttps://blog.schoergendorfer.com/post/reasoning-shortcuts-and-value-symmetries-what-symmetry-permits-architecture-real-2SecFeedde2026-08-25T09:01:49.726311+00:00Reasoning Shortcuts and Value Symmetries: What Symmetry Permits, Architecture Realizes, and Optimization Selectshttps://blog.schoergendorfer.com/post/hierarchical-compositionality-for-an-assistive-ai-agent-2SecFeedde2026-08-25T09:01:49.698699+00:00Hierarchical Compositionality for An Assistive AI Agenthttps://blog.schoergendorfer.com/post/the-greatness-of-science-cannot-be-planned-agentic-auto-research-is-fuzz-testingSecFeedde2026-08-25T09:01:49.672488+00:00The Greatness of Science Cannot Be Planned: Agentic Auto-Research is Fuzz Testinghttps://blog.schoergendorfer.com/post/time-present-and-time-past-benchmarking-large-language-models-on-temporally-evol-2SecFeedde2026-08-25T09:01:49.649704+00:00Time Present and Time Past: Benchmarking Large Language Models on Temporally Evolving Document Understandinghttps://blog.schoergendorfer.com/post/improving-auto-design-of-neural-pde-solvers-with-a-domain-specific-language-2SecFeedde2026-08-25T09:01:49.603835+00:00Improving Auto-Design of Neural PDE Solvers with a Domain-Specific Languagehttps://blog.schoergendorfer.com/post/nova-an-end-to-end-mlir-compiler-for-deep-learning-2SecFeedde2026-08-25T09:01:49.580526+00:00Nova: An End-to-End MLIR Compiler for Deep Learninghttps://blog.schoergendorfer.com/post/a-cross-architecture-audit-of-direction-based-inference-time-defences-in-visionSecFeedde2026-08-25T09:01:49.558916+00:00A Cross-Architecture Audit of Direction-Based Inference-Time Defences in Vision-Language Modelshttps://blog.schoergendorfer.com/post/crossing-the-margin-cliff-toward-relearn-robust-llm-unlearning-via-margin-calibrSecFeedde2026-08-25T09:01:49.538918+00:00Crossing the Margin Cliff: Toward Relearn-Robust LLM Unlearning via Margin Calibrationhttps://blog.schoergendorfer.com/post/rsmem-knowledge-enhanced-memory-evolution-for-remote-sensing-agents-with-systema-2SecFeedde2026-08-25T09:01:49.514173+00:00RSMeM: Knowledge-Enhanced Memory Evolution for Remote Sensing Agents with Systematic Evaluationhttps://blog.schoergendorfer.com/post/telco-gaia-bilingual-benchmark-for-agents-in-telecom-domain-2SecFeedde2026-08-25T09:01:49.490357+00:00Telco-GAIA: Bilingual Benchmark for Agents in Telecom Domainhttps://blog.schoergendorfer.com/post/cachespec-finding-the-sweet-spot-for-small-models-in-large-language-modelsSecFeedde2026-08-25T09:01:49.431913+00:00CacheSpec: Finding the Sweet Spot for Small Models in Large Language Modelshttps://blog.schoergendorfer.com/post/evidence-aware-mapreduce-for-forkable-compute-2SecFeedde2026-08-25T09:01:49.376536+00:00Evidence-Aware MapReduce for Forkable Computehttps://blog.schoergendorfer.com/post/polyuquest-verifiable-structure-aware-web-rag-over-heterogeneous-graphs-2SecFeedde2026-08-25T09:01:49.345455+00:00PolyUQuest: Verifiable Structure-Aware Web RAG over Heterogeneous Graphshttps://blog.schoergendorfer.com/post/the-clinicians-veto-navigating-trust-liability-and-uncertainty-in-autonomous-aiSecFeedde2026-08-25T09:01:49.295750+00:00The Clinician's Veto: Navigating Trust, Liability, and Uncertainty in Autonomous AI Prescribinghttps://blog.schoergendorfer.com/post/lagrange-an-open-vocabulary-energy-based-sparse-framework-for-generalized-end-toSecFeedde2026-08-25T09:01:49.267694+00:00Lagrange: An Open-Vocabulary, Energy-Based Sparse Framework for Generalized End-to-End Drivinghttps://blog.schoergendorfer.com/post/moco-ais-a-contrastive-learning-framework-for-similarity-computation-of-vessel-tSecFeedde2026-08-25T09:01:49.233792+00:00MoCo-AIS: A Contrastive Learning Framework for Similarity Computation of Vessel Trajectorieshttps://blog.schoergendorfer.com/post/finacumen-financial-multimodal-reasoning-via-self-evolving-experience-memory-harSecFeedde2026-08-25T08:46:37.582831+00:00FinAcumen: Financial Multimodal Reasoning via Self-Evolving Experience Memory Harnesshttps://blog.schoergendorfer.com/post/incumbent-advantage-brand-bias-and-cognitive-manipulation-dynamics-in-llm-recommSecFeedde2026-08-25T08:46:37.556456+00:00Incumbent Advantage: Brand Bias and Cognitive Manipulation Dynamics in LLM Recommendation Systemshttps://blog.schoergendorfer.com/post/self-evolving-scientific-agent-discovers-generalizable-physically-reasoned-fluidSecFeedde2026-08-25T08:46:37.490238+00:00Self-Evolving Scientific Agent Discovers Generalizable Physically-Reasoned Fluid Controlhttps://blog.schoergendorfer.com/post/how-small-can-you-go-lora-fine-tuning-270m-8b-models-for-merchant-information-exSecFeedde2026-08-25T08:46:37.460661+00:00How Small Can You Go? LoRA Fine-Tuning 270M-8B Models for Merchant Information Extraction in Financial Transactionshttps://blog.schoergendorfer.com/post/zero-knowledge-verification-for-frontier-ai-training-is-possibleSecFeedde2026-08-25T08:46:37.436301+00:00Zero knowledge verification for frontier AI training is possiblehttps://blog.schoergendorfer.com/post/expected-value-alignment-for-generative-reward-modeling-in-formal-mathematics-veSecFeedde2026-08-25T08:46:37.402730+00:00Expected Value Alignment for Generative Reward Modeling in Formal Mathematics Verificationhttps://blog.schoergendorfer.com/post/mindhelper-closed-loop-embodied-mental-state-reasoning-for-precision-interventioSecFeedde2026-08-25T08:46:37.369289+00:00MindHelper: Closed-Loop Embodied Mental-State Reasoning for Precision Interventionhttps://blog.schoergendorfer.com/post/latent-reward-steering-an-adaptive-inference-time-framework-that-implicitly-prom-2SecFeedde2026-08-25T08:46:37.335991+00:00Latent Reward Steering: An Adaptive Inference-Time Framework that Implicitly Promotes Cognitive Behaviors in Reasoning LLMshttps://blog.schoergendorfer.com/post/mosaic-modular-orchestration-for-structured-agentic-intelligence-and-compositionSecFeedde2026-08-25T08:46:37.304569+00:00MOSAIC: Modular Orchestration for Structured Agentic Intelligence and Compositionhttps://blog.schoergendorfer.com/post/designing-benchmarks-for-knowledge-workSecFeedde2026-08-25T08:46:37.275549+00:00Designing Benchmarks for Knowledge Workhttps://blog.schoergendorfer.com/post/gim-evaluating-models-via-tasks-that-integrate-multiple-cognitive-domainsSecFeedde2026-08-25T08:46:37.243722+00:00GIM: Evaluating models via tasks that integrate multiple cognitive domainshttps://blog.schoergendorfer.com/post/to-agents-a-multi-agent-ai-framework-for-subjective-preference-guided-topology-oSecFeedde2026-08-25T08:46:37.215156+00:00TO-Agents: A Multi-Agent AI Framework for Subjective Preference-Guided Topology Optimizationhttps://blog.schoergendorfer.com/post/cognifold-always-on-proactive-memory-via-cognitive-folding-2SecFeedde2026-08-25T08:46:37.185320+00:00CogniFold: Always-On Proactive Memory via Cognitive Foldinghttps://blog.schoergendorfer.com/post/finstar-towards-financial-reasoning-with-time-series-reasoning-modelsSecFeedde2026-08-25T08:46:37.132580+00:00FinSTaR: Towards Financial Reasoning with Time Series Reasoning Modelshttps://blog.schoergendorfer.com/post/revisiting-the-effectiveness-of-llm-pruning-for-test-time-scalingSecFeedde2026-08-25T08:46:37.110155+00:00Revisiting the Effectiveness of LLM Pruning for Test-Time Scalinghttps://blog.schoergendorfer.com/post/maven-t-reinforced-heterogeneous-distillation-for-real-time-multi-agent-trajectoSecFeedde2026-08-25T08:46:37.084038+00:00MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Predictionhttps://blog.schoergendorfer.com/post/activation-steering-for-aligned-open-ended-generation-without-sacrificing-cohereSecFeedde2026-08-25T08:31:27.504664+00:00Activation Steering for Aligned Open-ended Generation without Sacrificing Coherencehttps://blog.schoergendorfer.com/post/atp-bench-towards-agentic-tool-planning-for-mllm-interleaved-generationSecFeedde2026-08-25T08:31:27.446404+00:00ATP-Bench: Towards Agentic Tool Planning for MLLM Interleaved Generationhttps://blog.schoergendorfer.com/post/reinforcing-the-worlds-edge-a-continual-learning-problem-in-the-multi-agent-worlSecFeedde2026-08-25T08:31:27.416227+00:00Reinforcing the World's Edge: A Continual Learning Problem in the Multi-Agent-World Boundaryhttps://blog.schoergendorfer.com/post/skillnet-create-evaluate-and-connect-ai-skills-2SecFeedde2026-08-25T08:31:27.388053+00:00SkillNet: Create, Evaluate, and Connect AI Skillshttps://blog.schoergendorfer.com/post/training-proactive-and-personalized-llm-agentsSecFeedde2026-08-25T08:31:27.340156+00:00Training Proactive and Personalized LLM Agentshttps://blog.schoergendorfer.com/post/slidegen-collaborative-multimodal-agents-for-scientific-slide-generationSecFeedde2026-08-25T08:31:27.317724+00:00SlideGen: Collaborative Multimodal Agents for Scientific Slide Generationhttps://blog.schoergendorfer.com/post/adar-a-framework-for-equipping-llms-with-adaptive-reasoningSecFeedde2026-08-25T08:31:27.294810+00:00AdaR: A Framework for Equipping LLMs with Adaptive Reasoninghttps://blog.schoergendorfer.com/post/macd-multi-agent-clinical-diagnosis-with-self-learned-knowledge-for-llmSecFeedde2026-08-25T08:31:27.267830+00:00MACD: Multi-Agent Clinical Diagnosis with Self-Learned Knowledge for LLMhttps://blog.schoergendorfer.com/post/beyond-benchmarking-scenario-based-evaluation-of-large-language-models-for-persoSecFeedde2026-08-25T08:31:27.237624+00:00Beyond Benchmarking: Scenario-Based Evaluation of Large Language Models for Personalized Learninghttps://blog.schoergendorfer.com/post/beyond-benchmarks-llm-evaluation-with-an-anthropomorphic-and-lifecycle-orientedSecFeedde2026-08-25T08:31:27.209148+00:00Beyond Benchmarks: LLM Evaluation with an Anthropomorphic and Lifecycle-oriented Roadmaphttps://blog.schoergendorfer.com/post/benchmarking-retrieval-augmented-generation-strategies-for-large-language-modelSecFeedde2026-08-25T08:31:27.182179+00:00Benchmarking Retrieval-Augmented Generation Strategies for Large Language Model-Based Travel Mode Choice Predictionhttps://blog.schoergendorfer.com/post/neural-symbolic-reasoning-over-knowledge-graphs-a-survey-from-a-query-perspectivSecFeedde2026-08-25T08:31:27.133744+00:00Neural-Symbolic Reasoning over Knowledge Graphs: A Survey from a Query Perspectivehttps://blog.schoergendorfer.com/post/evaluating-large-language-models-for-automatic-analysis-of-teacher-simulationsSecFeedde2026-08-25T08:31:27.106714+00:00Evaluating Large Language Models for automatic analysis of teacher simulationshttps://blog.schoergendorfer.com/post/online-design-of-dynamic-networks-2SecFeedde2026-08-25T08:31:27.076252+00:00Online design of dynamic networkshttps://blog.schoergendorfer.com/post/memory-enhanced-neural-solvers-for-routing-problemsSecFeedde2026-08-25T08:31:27.048065+00:00Memory-Enhanced Neural Solvers for Routing Problemshttps://blog.schoergendorfer.com/post/a-survey-on-human-ai-collaboration-with-large-foundation-modelsSecFeedde2026-08-25T08:31:27.014799+00:00A Survey on Human-AI Collaboration with Large Foundation Modelshttps://blog.schoergendorfer.com/post/how-to-train-a-critic-stably-and-efficientlySecFeedde2026-08-25T08:31:26.982655+00:00How to Train a Critic Stably and Efficientlyhttps://blog.schoergendorfer.com/post/swe-refactor-bench-can-coding-agents-complete-a-long-horizon-whole-repository-stSecFeedde2026-08-25T08:16:31.078543+00:00SWE Refactor Bench: Can Coding Agents Complete a Long-Horizon, Whole-Repository Stack Migration?https://blog.schoergendorfer.com/post/eg-arsa-an-expert-grounded-open-model-for-visual-road-safety-auditing-in-low-resSecFeedde2026-08-25T08:16:31.050081+00:00EG-ARSA: An Expert-Grounded Open Model for Visual Road Safety Auditing in Low-Resource Settingshttps://blog.schoergendorfer.com/post/physics-constrained-deep-learning-model-for-contactless-blood-pressure-monitorinSecFeedde2026-08-25T08:16:31.016109+00:00Physics-Constrained Deep Learning Model for Contactless Blood Pressure Monitoring from Triaxial Bodyseismographyhttps://blog.schoergendorfer.com/post/convergeflow-language-flow-with-provable-convergence-to-token-embeddingsSecFeedde2026-08-25T08:16:30.979149+00:00ConvergeFlow: Language Flow with Provable Convergence to Token Embeddingshttps://blog.schoergendorfer.com/post/the-interaction-tax-when-communication-erases-diversity-in-multi-agent-teamsSecFeedde2026-08-25T08:16:30.944674+00:00The Interaction Tax: When Communication Erases Diversity in Multi-Agent Teamshttps://blog.schoergendorfer.com/post/adapter-based-few-shot-continual-learning-for-malicious-packet-recognitionSecFeedde2026-08-25T08:16:30.909556+00:00Adapter-Based Few-Shot Continual Learning for Malicious Packet Recognitionhttps://blog.schoergendorfer.com/post/when-names-cross-scripts-a-source-grounded-benchmark-for-historical-entity-reconSecFeedde2026-08-25T08:16:30.850858+00:00When Names Cross Scripts: A Source-Grounded Benchmark for Historical Entity Reconciliation in the Mongol Worldhttps://blog.schoergendorfer.com/post/act-with-intent-distilling-behavior-intent-for-vision-language-action-modelsSecFeedde2026-08-25T08:16:30.821409+00:00Act with Intent: Distilling Behavior Intent for Vision-Language-Action Modelshttps://blog.schoergendorfer.com/post/whats-the-catch-evaluating-temporal-consistency-in-vision-language-modelsSecFeedde2026-08-25T08:16:30.788584+00:00What's the Catch? Evaluating Temporal Consistency in Vision-Language Modelshttps://blog.schoergendorfer.com/post/injecmem-memory-injection-attack-on-llm-agent-memory-systemsSecFeedde2026-08-25T08:16:30.755185+00:00InjecMEM: Memory Injection Attack on LLM Agent Memory Systemshttps://blog.schoergendorfer.com/post/machine-learning-assisted-inverse-design-of-pixelated-mmwave-patch-antennasSecFeedde2026-08-25T08:16:30.699459+00:00Machine Learning Assisted Inverse Design of Pixelated mmWave Patch Antennashttps://blog.schoergendorfer.com/post/reward-free-continual-adaptation-for-resilient-space-robotsSecFeedde2026-08-25T08:16:30.667903+00:00Reward-Free Continual Adaptation for Resilient Space Robotshttps://blog.schoergendorfer.com/post/chebbooster-a-training-free-approach-for-efficient-diffusion-transformer-inferenSecFeedde2026-08-25T08:16:30.606041+00:00ChebBooster: A Training-Free Approach for Efficient Diffusion Transformer Inference via Chebyshev-Inspired Extrapolationhttps://blog.schoergendorfer.com/post/adaptive-item-based-collaborative-structures-via-noise-rescheduling-in-diffusionSecFeedde2026-08-25T08:16:30.575553+00:00Adaptive Item-based Collaborative Structures via Noise Rescheduling in Diffusion for Generative Recommendationhttps://blog.schoergendorfer.com/post/right-sizing-llm-agent-decomposition-in-vat-determination-a-pilot-controlled-sweSecFeedde2026-08-25T08:16:30.542585+00:00Right-Sizing LLM-Agent Decomposition in VAT Determination: A Pilot Controlled Sweephttps://blog.schoergendorfer.com/post/towards-a-densing-law-for-user-representation-learning-at-billion-scale-capacitySecFeedde2026-08-25T08:16:30.510670+00:00Towards a Densing Law for User Representation Learning at Billion-Scale Capacityhttps://blog.schoergendorfer.com/post/cross-domain-multi-task-data-to-text-generation-without-in-domain-training-dataSecFeedde2026-08-25T08:16:30.476145+00:00Cross-Domain, Multi-Task Data-to-Text Generation without In-Domain Training Datahttps://blog.schoergendorfer.com/post/adversarial-entropy-inflation-against-gumbel-based-inference-verificationSecFeedde2026-08-25T08:01:59.466289+00:00Adversarial Entropy Inflation Against Gumbel-Based Inference Verificationhttps://blog.schoergendorfer.com/post/the-emergence-of-relevance-through-axiomatic-attention-patterns-during-lora-fineSecFeedde2026-08-25T08:01:59.407177+00:00The Emergence of Relevance Through Axiomatic Attention Patterns During LoRA Fine-Tuninghttps://blog.schoergendorfer.com/post/thinking-beyond-videos-unifying-video-reasoning-and-deep-research-for-open-worldSecFeedde2026-08-25T08:01:59.376730+00:00Thinking Beyond Videos: Unifying Video Reasoning and Deep Research for Open-World Video Agentshttps://blog.schoergendorfer.com/post/mycelial-search-a-graph-structured-metaheuristic-for-continuous-optimisationSecFeedde2026-08-25T08:01:59.342672+00:00Mycelial Search: A Graph-Structured Metaheuristic for Continuous Optimisationhttps://blog.schoergendorfer.com/post/fides-a-concordance-protocol-for-llm-generated-trading-strategiesSecFeedde2026-08-25T08:01:59.313612+00:00FIDES: A Concordance Protocol for LLM-Generated Trading Strategieshttps://blog.schoergendorfer.com/post/how-much-regularization-survives-averaging-update-masking-in-federated-learningSecFeedde2026-08-25T08:01:59.233996+00:00How Much Regularization Survives Averaging? Update Masking in Federated Learninghttps://blog.schoergendorfer.com/post/e2s-pruner-progressive-two-stage-evidence-fusion-for-visual-token-pruning-in-visSecFeedde2026-08-25T08:01:59.205870+00:00E2S-Pruner: Progressive Two-Stage Evidence Fusion for Visual Token Pruning in Vision-Language Modelshttps://blog.schoergendorfer.com/post/multi-winner-voting-with-argumentative-ballotsSecFeedde2026-08-25T08:01:59.148892+00:00Multi-Winner Voting with Argumentative Ballotshttps://blog.schoergendorfer.com/post/credal-large-language-models-for-semantic-commitment-under-uncertaintySecFeedde2026-08-25T08:01:59.118048+00:00Credal Large Language Models for Semantic Commitment under Uncertaintyhttps://blog.schoergendorfer.com/post/retrieval-augmented-classification-of-environmental-mitigations-in-hydropower-liSecFeedde2026-08-25T08:01:59.087623+00:00Retrieval-Augmented Classification of Environmental Mitigations in Hydropower Licensing Documentshttps://blog.schoergendorfer.com/post/think-only-when-needed-prompt-authority-control-for-selective-slow-path-intervenSecFeedde2026-08-25T08:01:59.049684+00:00Think Only When Needed: Prompt-Authority Control for Selective Slow-Path Intervention in Vision-Language-Action Manipulationhttps://blog.schoergendorfer.com/post/benthicdino-physics-informed-self-distillation-for-view-invariant-side-scan-sonaSecFeedde2026-08-25T08:01:59.007843+00:00BenthicDINO: Physics-Informed Self-Distillation for View-Invariant Side-Scan Sonar Representationshttps://blog.schoergendorfer.com/post/ai-surrogate-modeling-for-real-time-tokamak-equilibrium-prediction-benchmarkingSecFeedde2026-08-25T08:01:58.977131+00:00AI Surrogate Modeling for Real-Time Tokamak Equilibrium Prediction: Benchmarking Neural Architectures and Validation on EXL-50Uhttps://blog.schoergendorfer.com/post/netconfarena-an-executable-benchmark-for-llm-agents-in-closed-loop-network-confiSecFeedde2026-08-25T08:01:58.946807+00:00NetConfArena: An Executable Benchmark for LLM Agents in Closed-Loop Network Configurationhttps://blog.schoergendorfer.com/post/counterfactual-transition-graphs-evaluating-cross-class-transition-qualitySecFeedde2026-08-25T08:01:58.913831+00:00Counterfactual Transition Graphs: Evaluating Cross-Class Transition Qualityhttps://blog.schoergendorfer.com/post/language-chain-in-alignment-cross-lingual-ranking-preference-optimizationSecFeedde2026-08-25T07:46:38.998415+00:00Language Chain in Alignment: Cross-Lingual Ranking Preference Optimizationhttps://blog.schoergendorfer.com/post/pointing-vla-typed-spatial-grounding-interfaces-for-vision-language-action-manipSecFeedde2026-08-25T07:46:38.977356+00:00Pointing-VLA: Typed Spatial Grounding Interfaces for Vision-Language-Action Manipulationhttps://blog.schoergendorfer.com/post/statistical-machine-translation-systems-of-english-pnar-language-pair-some-insigSecFeedde2026-08-25T07:46:38.957049+00:00Statistical Machine Translation Systems of English-Pnar Language Pair : Some Insights of the Emperical Studyhttps://blog.schoergendorfer.com/post/literarybigfive-author-personalized-text-generation-in-a-unified-interpretable-sSecFeedde2026-08-25T07:46:38.935495+00:00LITERARYBIGFIVE: Author-Personalized Text Generation in a Unified Interpretable Spacehttps://blog.schoergendorfer.com/post/demixpert-decomposed-response-modeling-with-gaussian-mixtures-for-ood-single-celSecFeedde2026-08-25T07:46:38.912601+00:00DeMixPert: Decomposed Response Modeling with Gaussian Mixtures for OOD Single-Cell Perturbation Predictionhttps://blog.schoergendorfer.com/post/molecular-llm-agents-from-architectural-design-to-scientific-autonomySecFeedde2026-08-25T07:46:38.890583+00:00Molecular LLM Agents: From Architectural Design to Scientific Autonomyhttps://blog.schoergendorfer.com/post/polychirp-multi-species-birdsong-classification-using-tinyml-on-low-power-acoustSecFeedde2026-08-25T07:46:38.867784+00:00PolyChirp: Multi-Species Birdsong Classification Using TinyML on Low-Power Acoustic Sensorshttps://blog.schoergendorfer.com/post/shaping-the-evolutionary-dynamics-of-robot-morphology-via-adaptive-control-learnSecFeedde2026-08-25T07:46:38.843612+00:00Shaping the Evolutionary Dynamics of Robot Morphology via Adaptive Control Learninghttps://blog.schoergendorfer.com/post/cultural-moment-benchmark-evaluating-video-cultural-reasoning-and-grounding-in-sSecFeedde2026-08-25T07:46:38.816826+00:00Cultural Moment Benchmark: Evaluating Video Cultural Reasoning and Grounding in Southeast Asiahttps://blog.schoergendorfer.com/post/beyond-verdicts-a-graph-based-analysis-of-human-and-llm-reasoning-in-scientificSecFeedde2026-08-25T07:46:38.792807+00:00Beyond Verdicts: A Graph-Based Analysis of Human and LLM Reasoning in Scientific Fact-Checkinghttps://blog.schoergendorfer.com/post/fedcc-towards-addressing-label-distribution-skews-in-distillation-based-federateSecFeedde2026-08-25T07:46:38.746697+00:00FedCC: Towards Addressing Label Distribution Skews in Distillation-Based Federated Learninghttps://blog.schoergendorfer.com/post/beyond-surface-cues-disentangling-sociocultural-signals-in-multilingual-llmsSecFeedde2026-08-25T07:46:38.720012+00:00Beyond Surface Cues: Disentangling Sociocultural Signals in Multilingual LLMshttps://blog.schoergendorfer.com/post/unlearning-is-not-just-erasing-temporal-decoupling-via-generation-inequalitySecFeedde2026-08-25T07:46:38.687218+00:00Unlearning Is Not Just Erasing: Temporal Decoupling via Generation Inequalityhttps://blog.schoergendorfer.com/post/splitlite-low-rank-residual-compression-for-split-learningSecFeedde2026-08-25T07:46:38.663886+00:00SplitLite: Low-Rank Residual Compression for Split Learninghttps://blog.schoergendorfer.com/post/coarse-indexing-fine-evidence-decoupling-temporal-granularity-in-long-video-ragSecFeedde2026-08-25T07:46:38.635396+00:00Coarse Indexing, Fine Evidence: Decoupling Temporal Granularity in Long-Video RAGhttps://blog.schoergendorfer.com/post/a-physical-response-and-memory-model-for-muon-optimizationSecFeedde2026-08-25T07:46:38.606837+00:00A Physical Response-and-Memory Model for Muon Optimizationhttps://blog.schoergendorfer.com/post/what-proves-you-wrong-benchmarking-language-models-on-falsifiable-research-ideatSecFeedde2026-08-25T07:46:38.557566+00:00What Proves You Wrong: Benchmarking Language Models on Falsifiable Research Ideationhttps://blog.schoergendorfer.com/post/wildhandbench-a-benchmark-for-handwritten-text-understanding-that-challenges-mllSecFeedde2026-08-25T07:46:38.531367+00:00WildHandBench: A Benchmark for Handwritten Text Understanding that Challenges MLLMs and Humanshttps://blog.schoergendorfer.com/post/deep-learning-based-multi-user-communication-design-for-dense-iot-networks-interSecFeedde2026-08-25T07:31:36.770693+00:00Deep Learning-Based Multi-User Communication Design for Dense IoT Networks: Interference-Aware Finite-Blocklength Communication and Preliminary MIMO Extensionshttps://blog.schoergendorfer.com/post/safety-hacking-in-constrained-best-of-n-inference-time-scalingSecFeedde2026-08-25T07:31:36.739626+00:00Safety Hacking in Constrained Best-of-$N$ Inference-time Scalinghttps://blog.schoergendorfer.com/post/aradetox-a-multi-dialect-arabic-detoxification-datasetSecFeedde2026-08-25T07:31:36.689731+00:00AraDetox: A Multi-Dialect Arabic Detoxification Datasethttps://blog.schoergendorfer.com/post/fairness-aware-mixture-of-experts-via-subgroup-reweighting-and-gate-regularizatiSecFeedde2026-08-25T07:31:36.586596+00:00Fairness-Aware Mixture-of-Experts via Subgroup Reweighting and Gate Regularizationhttps://blog.schoergendorfer.com/post/sdoh-aware-narrative-anchoring-bias-in-medical-llms-for-trustworthy-clinical-decSecFeedde2026-08-25T07:31:36.559382+00:00SDoH-Aware Narrative Anchoring Bias in Medical LLMs for Trustworthy Clinical Decision Supporthttps://blog.schoergendorfer.com/post/trace-a-self-evolving-skill-bank-for-consistent-limit-aware-llm-agentsSecFeedde2026-08-25T07:31:36.507571+00:00TRACE: A Self-Evolving Skill Bank for Consistent, Limit-Aware LLM Agentshttps://blog.schoergendorfer.com/post/dont-repeat-yourself-stopping-verbatim-loops-at-sampling-timeSecFeedde2026-08-25T07:31:36.474927+00:00Don't Repeat Yourself: Stopping Verbatim Loops at Sampling Timehttps://blog.schoergendorfer.com/post/xtc-head-aware-sampling-by-excluding-top-choicesSecFeedde2026-08-25T07:31:36.443181+00:00XTC: Head-Aware Sampling by Excluding Top Choiceshttps://blog.schoergendorfer.com/post/diarelay-relaying-dialogue-context-with-a-constant-size-memory-for-emotion-recogSecFeedde2026-08-25T07:31:36.414164+00:00DiaRelay: Relaying Dialogue Context with a Constant-Size Memory for Emotion Recognition in Conversationhttps://blog.schoergendorfer.com/post/object-uni-a-unified-model-for-object-centric-spatial-understanding-and-controllSecFeedde2026-08-25T07:31:36.389296+00:00Object-Uni: A Unified Model for Object-Centric Spatial Understanding and Controllable Generationhttps://blog.schoergendorfer.com/post/tee-x-tee-aware-acceleration-framework-for-large-vision-models-at-the-edgeSecFeedde2026-08-25T07:31:36.361746+00:00TEE-X: TEE-aware Acceleration Framework for Large Vision Models at the Edgehttps://blog.schoergendorfer.com/post/enrich-retrieve-rank-scaling-capability-discovery-beyond-in-context-routingSecFeedde2026-08-25T07:31:36.334851+00:00Enrich-Retrieve-Rank: Scaling Capability Discovery Beyond In-Context Routinghttps://blog.schoergendorfer.com/post/raco-reliability-aware-coarse-goal-optimization-for-inspection-oriented-uav-visiSecFeedde2026-08-25T07:31:36.303362+00:00RACO: Reliability-Aware Coarse-Goal Optimization for Inspection-Oriented UAV Vision-Language Navigationhttps://blog.schoergendorfer.com/post/hyperbolic-hierarchical-clustering-for-visual-representation-learningSecFeedde2026-08-25T07:31:36.271659+00:00Hyperbolic Hierarchical Clustering for Visual Representation Learninghttps://blog.schoergendorfer.com/post/physical-agentic-ai-an-architecture-for-orchestrating-a-robot-crew-with-llmsSecFeedde2026-08-25T07:31:36.238683+00:00Physical Agentic AI: An Architecture for Orchestrating a Robot Crew with LLMshttps://blog.schoergendorfer.com/post/evaluating-inference-time-defenses-against-package-hallucination-in-llm-generateSecFeedde2026-08-25T07:16:46.538060+00:00Evaluating Inference-Time Defenses Against Package Hallucination in LLM-Generated Codehttps://blog.schoergendorfer.com/post/do-not-copypaste-soft-barriers-for-copying-in-ai-assisted-programmingSecFeedde2026-08-25T07:16:46.476008+00:00Do Not Copy/Paste: Soft Barriers for Copying in AI-Assisted Programminghttps://blog.schoergendorfer.com/post/georisk-rag-a-hierarchy-aware-risk-framework-for-improving-rag-reliability-throuSecFeedde2026-08-25T07:16:46.441880+00:00GeoRisk-RAG: A Hierarchy-Aware Risk Framework for Improving RAG Reliability through Selective Answeringhttps://blog.schoergendorfer.com/post/teaching-llms-how-icu-physicians-approach-clinical-reasoning-through-omop-aligneSecFeedde2026-08-25T07:16:46.413241+00:00Teaching LLMs How ICU Physicians Approach Clinical Reasoning Through OMOP-Aligned Retrieval Improves Reasoning Across Clinical Domainshttps://blog.schoergendorfer.com/post/vision-language-models-for-occupational-physical-exposure-assessment-estimatingSecFeedde2026-08-25T07:16:46.353939+00:00Vision-Language Models for Occupational Physical Exposure Assessment: Estimating External Hand Forces in Manual Material Handling Tasks from RGB Videohttps://blog.schoergendorfer.com/post/clinical-graph-jepa-predictive-patient-state-knowledge-graphs-for-cognitive-deciSecFeedde2026-08-25T07:16:46.326126+00:00Clinical Graph-JEPA: Predictive Patient-State Knowledge Graphs for Cognitive Decision Supporthttps://blog.schoergendorfer.com/post/hybrid-panels-toward-human-ai-collaboration-in-survey-researchSecFeedde2026-08-25T07:16:46.295248+00:00Hybrid Panels: Toward Human-AI Collaboration in Survey Researchhttps://blog.schoergendorfer.com/post/functional-compatibility-as-a-determinant-of-persistent-neural-learningSecFeedde2026-08-25T07:16:46.245060+00:00Functional compatibility as a determinant of persistent neural learninghttps://blog.schoergendorfer.com/post/empire-explicit-manipulation-planning-as-a-learnable-intermediate-representationSecFeedde2026-08-25T07:16:46.215509+00:00EMPIRE: Explicit Manipulation Planning as a Learnable Intermediate Representation for Egocentric Hand-Motion Forecastinghttps://blog.schoergendorfer.com/post/rank-reversal-in-multilingual-llm-judges-a-label-free-double-centering-calibratoSecFeedde2026-08-25T07:16:46.187738+00:00Rank Reversal in Multilingual LLM Judges: A Label-Free Double-Centering Calibratorhttps://blog.schoergendorfer.com/post/cross-subject-generalization-in-decoding-perceived-speech-from-non-invasive-braiSecFeedde2026-08-25T07:16:46.158658+00:00Cross-Subject Generalization in Decoding Perceived Speech from Non-Invasive Brain Recordingshttps://blog.schoergendorfer.com/post/kontograph-verified-point-in-time-feature-consistency-and-amortised-explanationSecFeedde2026-08-25T07:16:46.129773+00:00KONTOGRAPH: Verified Point-in-Time Feature Consistency and Amortised Explanation for Real-Time Anti-Money Laundering under a 200 ms Decision Budgethttps://blog.schoergendorfer.com/post/probel-propaganda-detection-with-techniques-spans-and-explanationsSecFeedde2026-08-25T07:16:46.097084+00:00ProBel: Propaganda Detection with Techniques, Spans, and Explanationshttps://blog.schoergendorfer.com/post/self-supervised-graph-representation-learning-for-in-the-wild-wearable-and-smartSecFeedde2026-08-25T07:16:46.066309+00:00Self-Supervised Graph Representation Learning for In-The-Wild Wearable and Smartphone based Emotion Recognitionhttps://blog.schoergendorfer.com/post/pre-decoding-acoustic-triage-for-budgeted-vision-language-captioning-of-untrimmeSecFeedde2026-08-25T07:16:46.039820+00:00Pre-Decoding Acoustic Triage for Budgeted Vision-Language Captioning of Untrimmed Egocentric Videohttps://blog.schoergendorfer.com/post/sane-state-anomaly-neutralization-for-stable-extreme-context-delta-rule-modelsSecFeedde2026-08-25T07:16:46.011773+00:00SANE: State Anomaly Neutralization for Stable Extreme-Context Delta-Rule Modelshttps://blog.schoergendorfer.com/post/multimodal-examination-answer-data-with-expert-designed-outcome-based-educationSecFeedde2026-08-25T07:16:45.978412+00:00Multimodal examination answer data with expert-designed Outcome-Based Education rubrics for criterion-level assessmenthttps://blog.schoergendorfer.com/post/transhands-repurposing-human-pose-encoders-as-hand-pose-encodersSecFeedde2026-08-25T07:01:46.026442+00:00TransHands: Repurposing Human Pose Encoders as Hand Pose Encodershttps://blog.schoergendorfer.com/post/register-shifts-break-llm-safety-a-bengali-benchmark-with-culturally-grounded-haSecFeedde2026-08-25T07:01:46.001185+00:00Register Shifts Break LLM Safety: A Bengali Benchmark with Culturally Grounded Harmshttps://blog.schoergendorfer.com/post/llm-evaluation-on-unseen-questions-contextual-multidimensional-irt-modelSecFeedde2026-08-25T07:01:45.978533+00:00LLM Evaluation on Unseen Questions: Contextual Multidimensional IRT Modelhttps://blog.schoergendorfer.com/post/the-imitator-game-benchmarking-robot-imitative-ability-beyond-action-predictionSecFeedde2026-08-25T07:01:45.954070+00:00The Imitator Game: Benchmarking Robot Imitative Ability Beyond Action Predictionhttps://blog.schoergendorfer.com/post/learning-from-the-test-self-referential-differential-testing-for-deep-rl-agentsSecFeedde2026-08-25T07:01:45.930268+00:00Learning from the Test: Self-Referential Differential Testing for Deep RL Agents