Research papers — 2026-09-21

Today's research is dominated by new insights into the complex environments of neutron stars, ranging from their internal structures to their gravitational and electromagnetic signatures. In a wide parameter-space search using advanced detectors, researchers explored uncharted regions of continuous gravitational waves from unknown binary neutron stars. This study covered frequencies between 50 and 1000 Hz and orbital periods under three days.

While no signal was detected, the study sets new constraints on stellar asymmetry. It excludes neutron stars within 100 parsecs rotating faster than 495 Hz from having ellipticities above 5.2 times 10 to the power of negative eight. Complementing this work, a new orbital-free density functional theory method has been developed using a self-consistent extended Thomas-Fermi expansion. This method successfully modeled exotic nuclear pasta structures in the inner crust, such as connected rods and slashed with holes, without relying on empirical geometric assumptions.

Shifting from internal structures to cosmic events, studies of kilonovae suggest that uncertainties in the velocity distribution of lanthanide-rich ejecta can cause errors in mass estimation. Specifically, these uncertainties can cause factor of two to four errors when estimating ejecta mass from infrared peaks. Finally, testing the Efron-Petrosian method reveals it fails to recover the inverse-square distance law for radio pulsar fluxes when detection thresholds scale non-linearly with signal-to-noise ratios. This is a critical finding for pulsar population studies.

The transition from theoretical capability to reliable autonomy requires addressing how models manage their own internal states, particularly regarding uncertainty and self-reference. New causal evidence suggests that large language models do not merely output confidence scores but actively use them to drive behavior, such as deciding when to abstain from an answer. By using activation steering to boost or suppress confidence signals, researchers demonstrated that these signals directly control abstention rates.

This reveals that models deploy internal confidence representations alongside instructed thresholds to implement metacognitive policies. Such a capacity for structured control is central to the debate over recursive self-improvement. While theoretical frameworks like Kleene's Second Recursion Theorem suggest that introspective programs can exist, current transformer architectures face structural bottlenecks.

These bottlenecks include a lack of complete self-access and the feedforward nature of their processing, which prevent them from reaching a true introspection threshold. This gap between quasi-introspective metacognition and the fixed-point iteration required for sustainable self-evolution remains a primary architectural challenge.

The shift toward agentic systems is increasingly defined by how they communicate and structure their reasoning. Research into test-time communication shows that a team of k communicating agents can match the success rate of 4k independent agents on the ARC-AGI-3 benchmark. These gains compound as scale increases.

This collaborative advantage even allows teams to solve tasks that are impossible for any single agent. For example, they achieved superior results in polyomino packing and produced a 1,957-byte MNIST classifier that outperforms both human and single-agent benchmarks. In hardware design, moving away from direct RTL coding toward higher-level abstractions is proving similarly effective.

A workflow combining Agent-based HLS Design with RTL Refinement achieves a 2.6 times geometric-mean speedup over direct design. Meanwhile, in robotics, the KnowDemo framework uses vision-language models to extract task requirements from human videos. This allows robots to generate diverse demonstrations with alternative contact strategies rather than just mimicking motion.

These advancements suggest that the path to more capable autonomy lies in moving beyond isolated execution, whether through inter-agent dialogue or higher-level semantic reasoning. However, the challenge of high-dimensional continuous control remains a central tension in reinforcement learning. This is particularly true when the components of planning and learning become misaligned during training.

Learned sampling policies can diverge from planner behavior, and the distributions stored in replay buffers often become stale as models and value functions evolve. To address this, GEM-MPC introduces an MPPI-based method that integrates a policy trained to clone the planner with a KL-regularized policy designed to explore around it. This approach balances exploitation with guided exploration.

To mitigate the computational burden of refreshing stale planning data through full reanalysis, the framework employs Gated Prior Distillation. This method selectively learns from stored distributions only when they offer a superior target compared to the current prior. Across continuous-control benchmarks, this approach consistently outperforms existing planning-based baselines while operating under lower computational budgets.

Today's papers

• Wide parameter-space O3 search for continuous gravitational waves from unknown neutron stars in binary systems. "Vast regions of parameter space remain unexplored. [paper]
• The Effect of the Velocity Distribution on Kilonova Emission. This paper investigates how the velocity distribution of non-relativistic ejecta in neutron star mergers influences kilonova light-curves and spectra. [paper] [episode]
• Analog Gravity in Magneto-Viscous Fluids: Enhanced Analog Hawking Temperature in Accretion Disks. This paper investigates analog gravity within three-dimensional magnetohydrodynamic (MHD) flows, specifically focusing on how viscosity and magnetic field topology influence analog Hawking radiation in astrophysical accretion disks. [paper] [episode]
• Can the Efron-Petrosian Method Recover the Inverse-Square Distance Law for Simulated Radio Pulsar Fluxes?. This paper investigates whether the Efron-Petrosian (E-P) method, a statistical technique used to account for selection biases in astronomical datasets, can accurately recover the inverse-square law dependence of radio pulsar flux. [paper] [episode]
• Three-dimensional orbital-free density functional theory description of nuclear pasta in the inner crust of neutron stars. The authors propose an efficient method to calculate various nuclear pasta configurations in a non-empirical manner,... [paper] [episode]
• The Impact of Semantic Pairs on Self-Supervised Representation Learning. [paper]
• Beyond Reference-Based Evaluation: Reward Models for Meta-Evaluation of Grammatical Error Correction. [paper]
• Learning Gait-Aware Quadruped Locomotion with Temporal Logic Specifications. [paper]
• TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching. [paper]
• M2G-LLM: Enhancing Clinical Prediction via Multimodal Graph Reasoning and LLM Context Injection. [paper]
• LoRA Enhanced Contrastive Learning with SAS Vision Transformers. [paper]
• ExpBoN: Exponential-Noise Best-of- n for Efficient Test-Time LLM Alignment. [paper]
• HERMES: A Holistic End-to-End Risk-Aware Multimodal Embodied System with Vision-Language Models for Long-Tail Autonomous Driving. [paper]
• The Supersingular Isogeny Problem in Time and Memory p 1/3+o(1), Unconditionally. [paper]
• Benchmarking Autonomous Driving Planners Across Leaderboards: A Unified CARLA-Based Evaluation. [paper]
• IncentRL: The Trade-Off Between Preference Guidance and Task Performance. [paper]
• Decision-Focused Learning for Mean-Variance Portfolio Optimization via KKT-Based Reformulation. [paper]
• Rotational Mapping and Regional Atmospheric Retrievals of Variable Brown Dwarfs: Application to Luhman 16B and SIMP 0136. [paper]
• NetGent: Agent-Based Automation of Network Application Workflows. [paper]
• Deep Learning-Enhanced Real-Time Wi-Fi Sensing Through Single Transceiver Pair. [paper]
• Towards the Vision-Sound-Language-Action Paradigm: The HEAR Framework for Sound-Centric Manipulation. [paper]
• Toward accurate RUL and SoH estimation using reinforced graph-based physics-informed neural networks enhanced with dynamic weights. [paper]
• Understanding Structural Representation in Foundation Models for Polymers. [paper]
• How a Cooperative-Override Circuit Suppresses Nash Play in Large Language Models. [paper]
• Understanding LLM Quantization through Activation-Guided Compensation and Orthogonal Residuals. [paper]
• Diagnostic-Guided Longitudinal Modeling for Forecasting Retinal Atrophy Progression. [paper]
• The Weight Is Over - Interactive Diffusion on Consumer GPUs. [paper]
• Explanation-Bound Tool Execution for AI Agents: Server-Verified Action Claims Without Trusting Model Rationales. [paper]
• Self-Reference in Large Language Models: The Introspection Threshold for Recursive Self-Improvement. [paper]
• Bio-MF: Low-Latency and High-Fidelity EEG-to-fNIRS Cross-Modal Generation for Hybrid Motor-Imagery Brain--Computer Interfaces. [paper]
• Causal Evidence that Language Models use Confidence to Drive Behavior. [paper]
• The optical high-resolution spectral archive of the HERMES spectrograph mounted on the 1.2m Mercator telescope. [paper]
• ALMA Chemical Evolution (ACE) survey: an overview -- extending dust and gas inference to low metallicities at cosmic noon. [paper]
• Scaling Discovery through Test-Time Communication. [paper]
• KnowDemo: Knowledge-Guided Robot Demonstration Generation from Human Videos. [paper]
• OneBid: A Unified Auto-Bidding Foundation Model for Diverse oCPX Advertising Scenarios. [paper]
• Can Agents Design Better Chips with a Higher Level Abstraction?. [paper]
• Efficient Benchmarking in Production: A Study of an Evolving LLM Agent. [paper]
• Verify, Don't Trust: Agentic Model Development for Video Discovery Retrieval at Scale. [paper]
• Interference-Driven Clustered Optimisation for FM Spectrum Coordination. [paper]
• When Steering Fails in Latent Reasoning: A Latent-to-Language Transition Gap. [paper]
• MIST: Multimodal Survival Prediction with Genomic-Guided Histology Attention. [paper]
• Chinese Competitive Debating Dataset and Benchmark. [paper]
• A Neural Operator Emulator for Coastal and Riverine Shallow Water Dynamics. [paper]
• Listen Before You Speak: Response Planning from Listener Facial Reactions for Conversational Speech Generation. [paper]
• Rollout Total Correlation for Deep Reinforcement Learning. [paper]
• Optimal Scoring Rule Design under Partial Knowledge. [paper]
• EnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the Enterprise. [paper]
• Post-Rejection Follow-up Sampling: Measuring Outcomes of Rejected Decisions in Algorithmic DEX Trading. [paper]
• ISOMORPH: A Supply Chain Digital Twin for Simulation, Dataset Generation, and Forecasting Benchmarks. [paper]
• Tubular Neighbourhoods of Pfaffian Sets and Applications to Neural Networks. [paper]
• Toward Efficient Weakly Supervised Semantic Segmentation Using Only Low-Magnification Histopathological Images. [paper]
• Predictive Suppression Layers for Communication-Efficient Spiking Neural Networks. [paper]
• MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks. [paper]
• Training-Adaptive Convolutional Sparse Coding via Information Bottleneck for Robust Visual Representation. [paper]
• Elastic Threshold Attention: Learned Contextual Sparsity for Long-Context Decoding. [paper]
• ArenaFlow: From Trajectory Ranking to Hierarchical Credit Propagation for Open-Ended Agent RL. [paper]
• RegKT: Interpretable and Robust Deep Knowledge Tracing With IRT-Regularizer. [paper]
• Adaptive Rollout Truncation Based on Epistemic Uncertainty for Efficient Offline World Model Training. [paper]
• Hiding in Plain Sight: A Diffusion-based Mitigation of Geolocation Privacy Leakage in Vision-Language Models. [paper]

The papers

• Wide parameter-space O3 search for continuous gravitational waves from unknown neutron stars in binary systems — "Continuous gravitational waves, i.e., persistent and nearly-monochromatic signals emitted by asymmetric spinning neutron stars, remain elusive.
• Analog Gravity in Magneto-Viscous Fluids: Enhanced Analog Hawking Temperature in Accretion Disks — This paper investigates analog gravity within three-dimensional magnetohydrodynamic (MHD) flows, specifically focusing on how viscosity and magnetic field topology influence analog Hawking radiation in astrophysical accretion disks. [episode]
• The Effect of the Velocity Distribution on Kilonova Emission — This paper investigates how the velocity distribution of non-relativistic ejecta in neutron star mergers influences kilonova light-curves and spectra. [episode]
• Can the Efron-Petrosian Method Recover the Inverse-Square Distance Law for Simulated Radio Pulsar Fluxes? — This paper investigates whether the Efron-Petrosian (E-P) method, a statistical technique used to account for selection biases in astronomical datasets, can accurately recover the inverse-square law dependence of radio pulsar flux. [episode]
• Cross Subtype Transferability of Machine Learning Photometric Redshift Relations in Low Redshift Seyfert AGN — Based on your instructions-ing own ability-to-be-on-and-extingwisely-per of on own ability for the maning own ability’s ability to follow own own서ing서000 own dimming of own nuting never mind, never mind own서 diming of own서ng for ability vanish and extinction were never-
• Three-dimensional orbital-free density functional theory description of nuclear pasta in the inner crust of neutron stars — The authors propose an efficient method to calculate various nuclear pasta configurations in a non-empirical manner, without specifying the resulting geometric shapes a priori, based on "three-dimensional orbital-free density functional theory (OF-DFT)." The core of this approach [episode]
• Rollout Total Correlation for Deep Reinforcement Learning —
• Reinforcement Learning under External Influence: Guarantees, Algorithms, and Sample Complexity —
• Hierarchical attention interpretation: an interpretable speech-level transformer for bi-modal depression detection —
• Soda: An Object-Oriented Functional Language for Specifying Human-Centered Problems —
• Continuous Spiking Graph Neural Networks —
• Securing Large Language Models: Addressing Bias, Misinformation, and Prompt Attacks —
• Prevention is better than cure? Feedback from high specific energy winds in cosmological simulations with Arkenstone —
• dSTAR: Straggler Tolerant and Byzantine Resilient Distributed SGD —
• OverThink: Slowdown Attacks on Reasoning LLMs —
• OneTrack-M: A multitask approach to transformer-based MOT models —
• A Neural Operator Emulator for Coastal and Riverine Shallow Water Dynamics —
• First constraints on QCD axion dark matter using James Webb Space Telescope observations —
• Cultural Alignment in Large Language Models Using Soft Prompt Tuning —
• A Novel Relationship Between Gamma Ray Burst Duration And Photospheric Radius —
• Trajectory Entropy Reinforcement Learning for Robust Robot Motor Skill Learning —
• Understanding In-context Learning of Addition via Activation Subspaces —
• SRAF: Stealthy and Robust Adversarial Fingerprint for Copyright Verification of Large Language Models —
• VQ-Logits: Compressing the Output Bottleneck of Large Language Models via Vector Quantized Logits —
• Millimeter-wave observations of Euclid Deep Field South using the South Pole Telescope: A data release of temperature maps and catalogs —
• CASE: Contrastive Activation for Class-Sensitive Explanations —
• Cosmographic constraints from late-time probes including fast radio bursts —
• Toward accurate RUL and SoH estimation using reinforced graph-based physics-informed neural networks enhanced with dynamic weights —
• Collab-Solver: Collaborative Solving Policy Learning for Mixed-Integer Linear Programming —
• GWTC-4.0: Population Properties of Merging Compact Binaries —
• Are Companies Taking AI Risks Seriously? A Systematic Analysis of Companies' AI Risk Disclosures in SEC 10-K forms —
• NetGent: Agent-Based Automation of Network Application Workflows —
• Benchmark of stylistic variation in LLM-generated texts —
• Generalizing Beyond Suboptimality: Offline Reinforcement Learning Learns Effective Scheduling through Random Solutions —
• Robust Mixture Models for Algorithmic Fairness Under Latent Heterogeneity —
• A regret minimization approach to fixed-point iterations —
• Benchmarking Autonomous Driving Planners Across Leaderboards: A Unified CARLA-Based Evaluation —
• Fidel-TS: A High-Fidelity Multimodal Benchmark for Time Series Forecasting —
• Fact Grounded Attention: Eliminating Hallucination in Large Language Models Through Attention Level Knowledge Integration —
• Transformers Discover Molecular Structure Without Graph Priors —
• Auditing a KB Elicitation of Frontier LLM Knowledge: A Multi-dimensional Analysis of GPTKB v1.5 —
• MIRANDA: short signatures from a leakage-free full-domain-hash scheme —
• The Impact of Semantic Pairs on Self-Supervised Representation Learning —
• The Blue Jay Survey: Deep JWST Spectroscopy for a Representative Sample of Galaxies at Cosmic Noon —
• Provably Optimal Reinforcement Learning under Safety Filtering —
• PRIVET: PRoximIty leakage detection Via Extreme value Theory —
• Deep Learning-Enhanced Real-Time Wi-Fi Sensing Through Single Transceiver Pair —
• Probing the dawn of galaxies: star formation and feedback in the JWST era through the GAEA model —
• Nearly forgotten results in development of physical cosmology —
• Souper-Model: How Simple Arithmetic Unlocks State-of-the-Art LLM Performance —
• SteganoBackdoor: Evading Data-Poisoning Defenses via Steganographic Backdoors —
• A Hybrid Computational Intelligence Framework for scRNA-seq Imputation: Integrating scRecover and Random Forests —
• K2-V2: A 360-Open, Reasoning-Enhanced LLM —
• Spatially Resolved Physical Properties of Young Star Clusters and Star-forming Clumps in the Brightest z>6 Galaxy, the Strongly Lensed Cosmic Spear at z=6.2 —
• Measuring the neutron star equation of state from EMRIs in dark matter environments with LISA —
• Understanding Structural Representation in Foundation Models for Polymers —
• Boltzmann generators for amorphous particle systems —
• Foundations and Design Principles of Lightweight Cryptography for IoT Systems —
• Overmassive black holes and little red dots naturally form in simulations —
• Explainable Multimodal Aspect-Based Sentiment Analysis with Dependency-guided Large Language Model —
• BEAT-Net: Injecting Biomimetic Spatio-Temporal Priors for Interpretable ECG Diagnosis —
• MemeLens: Multilingual Multitask VLMs for Memes —
• Impact of soft versus hard equations of state on hybrid stars —
• Cross-Country Learning for National Infectious Disease Forecasting Using European Data —
• Learning to Advect: A Neural Semi-Lagrangian Architecture for Weather Forecasting —
• CORDS: Continuous Representations of Discrete Structures —
• Semantic Calibration Prevails Where Token Confidence Fails: Benchmarking Long-Form Scientific QA —
• HERMES: A Holistic End-to-End Risk-Aware Multimodal Embodied System with Vision-Language Models for Long-Tail Autonomous Driving —
• Near-Universal Multiplicative Updates for Nonnegative Einsum Factorization —
• MENASpeechBank: A Reference Voice Bank with Persona-Conditioned Multi-Turn Conversations for AudioLLMs —
• Gradient-Stable Attention Heads Signal LLM Correctness —
• MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks —
• Position: A Dynamical Systems Perspective is Needed to Advance Time Series Modeling —
• StableAML: Machine Learning for Behavioral Wallet Detection in Stablecoin Anti-Money Laundering on Ethereum —
• Neural ensemble Kalman filter: Data assimilation for compressible flows with shocks —
• The MAMA-MIA Challenge: Advancing Generalizability and Fairness in Breast MRI Tumor Segmentation and Treatment Response Prediction —
• Data-Driven Integration Kernels for Interpretable Nonlocal Operator Learning —
• Taming the Adversary: A Cost-to-Disturbance Ratio Approach to Adversarial Reinforcement Learning —
• Beyond Final Answers: CRYSTAL Benchmark for Transparent Multimodal Reasoning Evaluation —
• Towards the Vision-Sound-Language-Action Paradigm: The HEAR Framework for Sound-Centric Manipulation —
• How do LLMs Compute Verbal Confidence —
• Causal Evidence that Language Models use Confidence to Drive Behavior —
• Near-Optimal Primal-Dual Algorithm for Learning Linear Mixture CMDPs with Adversarial Rewards —
• Transferable knowledge graphs with executable learned operators for algorithm design —
• Constraining The Neutrino-Nucleon Cross Section with the Ultrahigh-Energy KM3NeT Event —
• Learning Surrogate LPV State-Space Models with Uncertainty Quantification —
• Nonnegative Matrix Factorization in the Component-Wise L1 Norm for Sparse Data —
• Offline Constrained RLHF with Multiple Preference Oracles —
• Is your AI Model Accurate Enough? The Difficult Choices Behind Rigorous AI Development and the EU AI Act —
• EAGLE: Edge-Aware Graph Learning for Proactive Delivery Delay Prediction in Smart Logistics Networks —
• Wild is the wind from low-luminosity AGN: A jet-driven gas bubble blowing out a massive CO-dark outflow in ESO 420-G13 —
• Amortized Filtering and Smoothing with Conditional Normalizing Flows —
• Evolving Skill Modules under a Fixed Planner: Versioning, Rollback, and Runtime Governance for Long-Lived Robot Systems —
• Lessons Without Borders? Evaluating Cultural Alignment of LLMs Using Multilingual Story Moral Generation —
• Stability Enhanced Gaussian Process Variational Autoencoders —
• Galactic Archaeology with the Subaru `=Onohi`ula Prime Focus Spectrograph Strategic Program —
• Diagnostic-Guided Longitudinal Modeling for Forecasting Retinal Atrophy Progression —
• LLM Safety From Within: Detecting Harmful Content with Internal Representations —
• JudgeSense: A Benchmark for Prompt Sensitivity in LLM-as-a-Judge Systems —
• How a Cooperative-Override Circuit Suppresses Nash Play in Large Language Models —
• Why Do LLMs Struggle in Strategic Play? Broken Links Between Observations, Beliefs, and Actions —
• REALM: An RGB- and Event-Aligned Latent Manifold for Cross-Modal Perception —
• Rhamba: Region-Aware Hybrid Attention-Mamba Framework for Self-Supervised Learning in Resting-State fMRI —
• Constraint Decay: The Fragility of LLM Agents in Backend Code Generation —
• BoostAPR: Boosting Automated Program Repair via Execution-Grounded Reinforcement Learning with Dual Reward Models —
• LiteMedCoT-VL: Parameter-Efficient Adaptation for Medical Visual Question Answering —
• On the Limitations of Large Language Models for Conceptual Database Modeling —
• The critical slowing down in training diffusion models —
• ISOMORPH: A Supply Chain Digital Twin for Simulation, Dataset Generation, and Forecasting Benchmarks —
• Exemplar Partitioning for Mechanistic Interpretability —
• Sometin Beta Pass Notin: Improving Multilingual ASR for Nigerian Languages via Knowledge Distillation —
• Correlation between baryonic process and galaxy assembly bias —
• Gravitational-wave constraints on H 0 are robust to putative redshift evolution in the binary black hole mass spectrum at current sensitivity —
• Draft-OPD: On-Policy Distillation for Speculative Draft Models —
• Survival Reinforcement Learning: Toward Scalable Self-Supervised RL —
• PaCo-VLA: Passivity-Shielded Compliance Prior for Contact-Rich Vision-Language-Action Manipulation —
• AgenticRL: Agentic Reinforcement Learning with Self-Refinement for Complex UAV Navigation —
• Scaling Novel Graph Generation via Lightweight Structure-Guided Autoregressive Models —
• Post-Rejection Follow-up Sampling: Measuring Outcomes of Rejected Decisions in Algorithmic DEX Trading —
• Geometry-Aware Reinforcement Learning for 2D Irregular Nesting —
• Observing Cosmic Reheating with the expanded Simons Observatory —
• Intent-Governed Tool Authorization for AI Agents —
• MemAudit: Auditing Long-Term Agent Memory via Hidden User-State Recovery —
• Recall Before Rerank: Benchmarking Deep Learning Models for Large-Scale Code-to-Code Retrieval —
• Learning Gait-Aware Quadruped Locomotion with Temporal Logic Specifications —
• The Binary Tree Mechanism is Optimal for Differentially Private Continual Counting —
• Chameleon: Recovering Cyber-Physical Systems from Memory Corruption Attacks via ML Surrogates —
• GeoSelect: Spatial-Program Execution for Training-Free Referring Remote Sensing Image Segmentation —
• Self-Reference in Large Language Models: The Introspection Threshold for Recursive Self-Improvement —
• Tubular Neighbourhoods of Pfaffian Sets and Applications to Neural Networks —
• Git-Assistant: Planning-Based Support for Updating Git Repositories —
• Toward Efficient Weakly Supervised Semantic Segmentation Using Only Low-Magnification Histopathological Images —
• Cover First, Disagree Softly: Rethinking Mismatch-First Active Learning for Frame-Level Audio Classification —
• TVGL-CFM:Generating and Forecasting Time-Varying Trajectories of Dynamic Networks with Conditional Flow Matching —
• ANNLib: A Development Framework for Efficient Approximate Nearest Neighbor Search —
• Teaching LLMs to Self-Evolve: Cultivating Core Meta-Skills with Reinforcement Learning —
• Decision trees, Frobenius traces, and Weierstrass coefficients of elliptic curves —
• Explanation-Bound Tool Execution for AI Agents: Server-Verified Action Claims Without Trusting Model Rationales —
• Contrastive Concept Importance: Explaining Pairwise Class Decisions Through Automatically Extracted Concept Representations —
• Convex losses and their applications to SVM, SVR, and Shallow Neural Networks —
• Wiktionary as a Crowdsourced Lexicon for English Dialects —
• Multi-turn Conversational AI from Text to Multimodal Interaction: Data, Models, Evaluation, and Open Challenges —
• Quasi-periodic Eruptions from Recurrent Satellite Black Hole Transits through Magnetized Galactic Nucleus Accretion Disks —
• Training-Adaptive Convolutional Sparse Coding via Information Bottleneck for Robust Visual Representation —
• Do small language models know what they don't know? —
• HERMES: Contrast-Aware Knowledge Graph Reasoning from Clinical Notes for Patient Outcome Prediction —
• TALON: A Temporally Aware Longitudinal Framework for Radiology Report Generation —
• From Discharge Notes to Patient Understanding: Persona-Grounded, Open-Ended Simulation of LLMs as Discharge Educators —
• Beyond WER: Entity and Disfluency Recall in Accented Conversational ASR —
• SAGE: Schema-Guided LLMs for Grant Review —
• Reviser: Revision-Capable Text Generation via Autoregressive Cursor Actions —
• Recursive Language Models Generalize Out of Domain —
• TatBLiMP: A Benchmark of Linguistic Minimal Pairs for Tatar —
• Transsion's Speaker-Attributed Multilingual ASR System for the MLC-SLM 2026 Challenge —
• Towards Secure Cloud-Native Computing: Unveiling Kubernetes Misconfigurations with Large Language Models —
• A Generative Grammar Underlying the Voynich Manuscript, the Pastiche Hypothesis: Evidence from Large Language Models —
• PhysioBench: A Unified Benchmark for Physiological Signal Question Answering —
• From Generation to Detection: Exploration of Discourse Driven Scenario based LLM Generated Fake News —
• Curriculum-Based Noise Adaptation for Phoneme-to-Text Reconstruction in Visual Speech Recognition —
• COAL-SQL: Coverage-Guided Augmentation and Failure-Driven Learning for Text-to-SQL Post-Training —
• VISPATH: Visual-Intent-Guided Path Reasoning for Multimodal Knowledge Graph Question Answering —
• Boosting Deepresearch and LongContext Ability with Self-Generated Deepresearch Rollouts Traces —
• Reading Less While Writing: A Closed-Form Bandwidth Dial for Streaming Multimodal Decoders —
• Rewarding Efficient Reasoning Improves Abstention on Underspecified Tasks in Reasoning Models —
• Reading Anxiety or Reading the Label? Comparing Fine-Tuned and Frontier Models for Anxiety Detection on Social Media —
• Enhancing Audio Reasoning via Semantic Summary Prediction —
• MME-Safety: A Fine-grained Benchmark for Safety Evaluation of MLLMs —
• Reconstruction of 4D Mitral Regurgitation Hemodynamics from Sparse Planar Data using Deep Operator Networks with Test-Time Adaptation —
• Automated Physics-Informed Neural-Networks-Based Calibration of Highly Segmented Silicon Telescopes —
• The Refutation Gap: Certifying Both Halves of an Optimality Claim —
• Cross-Lingual Parkinson's Disease Severity Assessment Using Pre-trained Speech Embeddings: A Multi-Class Evaluation —
• Reinforcement learning for post-coronagraphic wavefront control —
• Sparse Priors for Efficient Distribution Learning —
• The Right Tool for the Job: On the Selection of Mitigations for GenAI Privacy Threats —
• BI-Agent and BI-Bench: Towards Automating End-to-End Business Intelligence —
• Elastic Threshold Attention: Learned Contextual Sparsity for Long-Context Decoding —
• Extreme classification: beating chance with one training example from each class —
• SpaceDiffusion: Over-the-Orbit Diffusion for Space Generate-and-Forward Communications —
• Generative Artificial Intelligence Chatbots for Motivational Interviewing: A Scoping Review From System Design to Intervention Outcomes —
• Bio-MF: Low-Latency and High-Fidelity EEG-to-fNIRS Cross-Modal Generation for Hybrid Motor-Imagery Brain--Computer Interfaces —
• Continuous Delayed-Memory Stochastic Gradient Descent and Continuous-Time Reinforcement Learning from History of Astrophysical Time Series Studies —
• TPM-Attest: Hardware-Rooted Integrity Attestation as a Kernel-Level Anti-Cheat Alternative for Linux —
• Do Quantum Models Scale Like LLMs? —
• Free-Free Radio Emission from Little Red Dots as a Probe of Ionized Gas —
• Little Red Dots As Super-Eddington Fountain Flows —
• ALMA Chemical Evolution (ACE) survey: The gas fundamental metallicity relation at cosmic noon —
• Recovering the Morning and Evening Limbs of WASP-94Ab from Its Limb-Averaged JWST Transmission Spectrum —
• Cosmological Correlators from Resurgence —
• ALMA Chemical Evolution (ACE) survey: dust-to-gas ratios in sub-solar metallicity galaxies at cosmic noon —
• The Treasury of Extremely Metal-Poor O Stars —
• From IFS Maps to 3D Structure I: Constraining Geometry in the Circumgalactic Medium —
• The Galactic Phosphorus Survey. I. Chemical Evolution across the Galactic Disk from about 750 FGK Stars —
• Quiescent Galaxies at z > 2 from CAPERS spectroscopy and their Number densities —
• Cause of chromospheric opposite polarity intrusions discovered in Sunrise III/SCIP data: MURaM-ChE simulations point to twisted flux ropes —
• An Archival Calibration of JWST/MIRI Prism Wide-Field Slitless Spectroscopy: Methodology, Performance, and a Mid-Infrared Spectral Atlas of Galaxies at z=0-4 in the GOODS-N/S Fields —
• AutoTherm: Automated Thermal Field Theory rates for cosmology —
• Thermal gravitino and axino rate from AutoTherm —
• Cosmic web Ly alpha emission in a sample of overdense regions —
• White Dwarf Natal Kicks as Velocity-Space Random Walks. I. Fokker-Planck Theory —
• White Dwarf Natal Kicks as Velocity-Space Random Walks. II. Binary Stellar Evolution and Observational Kick Constraints —
• When AI Reviews Train AI Reviewers: Scientific-Judgment Collapse and Mitigation —
• Systematic Uncertainties and Their Impact on Gamma-Ray Searches for Dark Matter in Dwarf Galaxies —
• mu squared-Bench: A Multilingual Machine Unlearning Benchmark —
• Euclid Quick Data Release (Q1) Euclid spectroscopy of quasars. 2. Physical properties from spectral fitting —
• Efficient Bayes-Adaptive Reinforcement Learning with Temporal Logic Specifications —
• Sown at Cosmic Dawn: Searching for Pop III Signatures in JWST JADES He II Selected Line Emitters —
• From Switching to Dynamic Regret: A Simple Reduction via Unbiased Random Sequences —
• RBS-Attention: Radius-Bounded Sparse Prefill for Long-Context Large Language Models —
• Complex Problem Solving in Large Language Models: A Statistical Control Survey and Diagnostic Framework —
• Attention-Aware Routing: Coupling Routing and Attention in MoEs —
• Comparing the Detectability of the Baryon Acoustic Oscillation through Gravitational-Wave Autocorrelation and Cross-Correlation with Galaxies —
• Ranking Competing geologic interpretations via foundation-model-assisted generative hydrologic inversion —
• CaLR: Causal Latent Revision for Robust Diffusion Reasoning —
• ASGARD: Action-Space Guard for UAV Resilience via Reinforcement Learning —
• Trustworthy FinAInce: Unpacking How AI-Mediated Financial Advice is Judged —
• From Stress to Affect: Multimodal Deep Learning for Physiological Emotion Recognition Across Wearable Sensor Modalities —
• Voice-Light: A Full-Duplex Cascaded Voice Agent with Causal Turn-Taking and Speculative Generation —
• A fifth companion in the HR 8799 system revealed by Gaia —
• MOSAIC-SR: Transformer-Guided Symbolic Regression for Scientific Equation Recovery —
• Aggregated Posterior Predictive Checks for Generative Modeling —
• On the Limits of Maximal Coding Rate Reduction for Out-of-Distribution Generalisation —
• What Drives Galaxy-Galaxy Differences In The Selective Attenuation Curve? New Perspectives from DESI DR1 —
• Endothermic Z'-Portal Dark Matter: LZ-LHC Complementarity —
• Fragment-Aware Vision Transformers for Fresco-Fragment Style Classification —
• A Smoothed Discrepancy Principle for Random Feature Methods and Neural Networks —
• (Don't) Trust, but (Don't) Verify: Developers' Attention to Security in AI-Generated Code —
• The Galactic Dynamics of Free-Floating Planets: From Ejection Kicks to Microlensing —
• Scaling Discovery through Test-Time Communication —
• Self-Similar Mass Spectra of Hierarchical Black Hole Mergers —
• Stiefel-AdamW: Geometry-Aware AdamW for Linear Factorization Blocks —
• ALMA Chemical Evolution (ACE) survey: the dust content of subsolar metallicity galaxies at cosmic noon —
• A Lightweight Plug-in Gate for Transformer-Based Time-Series Forecasters —
• Physically Based Rendering in the Latent Space —
• FedeRage: Provably Convergent Agnostic Federated Learning under General Client Drift —
• How Much of a Real Workload Can LLM-Generated GPU Kernels Actually Reach? —
• PlantShade: Predicting Plant Shadows for Lighting-Aware Robotic Agricultural Operation —
• LoRA Enhanced Contrastive Learning with SAS Vision Transformers —
• ALMA Chemical Evolution (ACE) Survey: Molecular gas properties of low-mass, low-metallicity galaxies at cosmic noon —
• Toward individual-level calibration in affect recognition with perceptual adjustment queries —
• Aligning with Lived Experience: Heterogeneous Benefits of Fine Tuning in Mental Health Support Generation —
• A mechanism for fast radio bursts and their narrow spectra —
• Loopjacking: Hijacking Human-in-the-Loop Approval —
• The Hidden Cost of Digits: Number Normalization and WER in ASR Systems —
• Triply-Scalable Equivariant Gaussian Process Modeling —
• Tidal Dissipation from Circularization in Kepler Binaries —
• Origin Is All You Need: Provenance-Aware Transformers for Structural Trust-Boundary Separation —
• Geometry of Values: Task Vector Composition for Ethical Preference Alignment in Language Models —
• Detecting Hallucination in LLMs: Tracing the Topological Signatures of Impaired Context Sharing —
• NetInspector: Measuring and Improving LLM Capabilities for Reliable Intent-Based Networking Policy Generation —
• The Auger Radio Infill SKALA Extension (ARISE): First Air-Shower Measurements —
• REFINEPPO: Learning Continuous Control Policies by Iterative Action Refinement —
• Talk to Me, Jarvis: An Open-Source Edge-Deployable Voice Assistant Framework for Autonomous Racecars —
• Decoupling Internal Representational Changes and Causal Importance in Fine-Tuned Large Language Models —
• From Task Success to Productive Success: Evaluating Human-AI Collaboration by Quality and Cost —
• Energy Partition in Relativistic Electron-Positron-Ion Reconnection —
• Non-Conservative Mechanics in the Compact-Body Mass-Shell Model and its Waveform Structuring —
• Signal-Centric Remote Sensing via Alternative Preprocessing and Acoustic Processing for ML-Driven Applications —
• The Birth of Minor-Body Populations in the beta Pictoris System I: The First 25 Myr —
• Layerwise Decoupling for Stable Structured Sparsification of Fully Connected Layers —
• The Stochastic Shift: A New Evaluation Paradigm for Text-to-SQL with AI Operators —
• Granular Structure and Local de Broglie Wavelength in Fuzzy Dark Matter Halos —
• Effects of Physical Cooling on the Structure of Circumbinary Disks —
• Diverse and Adaptable Arm Coordination for Octopus-Crawling via Diffusion-Based Uncertainty-Aware Optimization —
• TinyCeNN-LM: Quality-Gated Conversion of Pretrained Attention with CeNN-Inspired Cellular-Recurrent Layers —
• Scaling Forced Alignment to End-User Devices —
• Toss If Perishable: An Ethnographic Study on Building Scenario-Based Training for Non-Perishable Skills —
• Clinician-Grounded Quality Assurance for AI-Assisted Psychiatric Intake —
• COPAS: Compact Objects and the Physics of Accretion Survey I. Accreting Binaries with P < 83 Minutes —
• EnSol: an environment-aware graph neural network for molecular solubility prediction —
• CoLearn: An Agentic Tutor that Learns its Learner in a Human--AI Co-Learning Loop —
• Can Agents Design Better Chips with a Higher Level Abstraction? —
• HMB-GAN: Hybrid Multi-B'ezier GAN for Vector Shape Synthesis —
• Repulsive normalizing flow mixtures for adaptive importance sampling: reliability analysis of complex systems —
• M2G-LLM: Enhancing Clinical Prediction via Multimodal Graph Reasoning and LLM Context Injection —
• SpecOpt: Contact-Diff Reasoning for Agentic Molecule Optimization Toward Binding Specificity —
• TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching —
• CRAFT HTR2: Polarimetry of 64 non-repeating fast radio bursts from the updated CRAFT catalogue —
• Not All Irregularity Is Equal: Causally Isolating a Rare Failure Mode in Japanese Morphological Inflection —
• Implicit Rule Induction with Test-Time Task Embeddings in ARC-like Tasks —
• I'll Keep an Ear Out: Teaching AudioLLMs Proactive Audio Assistance —
• When Better Turns Do Not Make Better Agents: Diagnosing the Gap Between Next-Turn Metrics and Workflow Success —
• SWE-Proof: Can Language Models Resolve Real-World Issues with Machine-Checked Proofs? —
• AI-GRACE: A Use-Case Operationalization Framework for Agentic AI: From Organizational Objectives and Obligations to Deployment Capabilities and Architecture —
• Reliability-Centered Evaluation of Sparse Longitudinal CT Lesion-Size Forecasting with Conformal Interval Calibration and Gompertz-Inspired Regularization —
• Information-Gain Rewards over Diversity-Pruned Tests: GT-Anchored Verifier Co-Training for Reliable Code Generation —
• Visual Navigation Transformer with Pose Attention —
• Ability-Residual Decoupled Modeling for Affective Cognitive Diagnosis —
• Fewer Steps, Better Actions: Rethinking Flow-Matching Inference for VLA Policies —
• X-SPUR: Explainable Surprisal-Based Protocol-Aware Unsupervised Reasoning for Automotive Ethernet Intrusion Detection —
• A Fully Differentiable Neuro-Soft-Symbolic Framework for Perceptual Task Planning —
• Combining Exploratory Analysis and Automated Analysis for Anomaly Detection in Real-Time Data Streams —
• Hallucination-R1: Robustness-Oriented Paraphrase Generation for Factual Consistency —
• FOCAL-VLA: Subtask-Guided Geometry Distillation and Implicit World Modeling for Vision-Language-Action Models —
• KnowDemo: Knowledge-Guided Robot Demonstration Generation from Human Videos —
• Beyond Reference-Based Evaluation: Reward Models for Meta-Evaluation of Grammatical Error Correction —
• Multiclass Semantic Segmentation of Wildland Fire Images Using Context-Aware Centralized Copy-Paste Data Augmentation —
• From Trainability Diagnostics to Optimization Claims: Boundaries and Controls in Variational Quantum Optimization —
• VLA-Scope: Shift-Aware Failure Prediction for Vision-Language-Action Models —
• When Does Reasoning Help in Machine Translation? A Hierarchical Analysis of LRM Reasoning Traces —
• Verify, Don't Trust: Agentic Model Development for Video Discovery Retrieval at Scale —
• CogGym: Towards Large-Scale Comparative Evaluation of Human and Machine Cognition —
• PlaceReasoner-Beta: Reasoning-Driven Macro Placement and Benchmarking —
• Programming AMD XDNA NPUs with Open-source Compiler Tools: A FlashAttention Case Study —
• Efficient Benchmarking in Production: A Study of an Evolving LLM Agent —
• Transcript-Bound Combiners for Downgrade-Resilient Hybrid Post-Quantum Key Establishment: Definition, Proof, and Embedded-Device Cost —
• Beyond Exact Match: Task-Aware GRPO for Cross-Domain PCBA Visual Question Answering —
• How Many Humans Are 32 LLM Judges Worth? —
• Neutrino Emission from Proton Photon Jet Interactions in Microquasars —
• MIRCID: Inferred Hub-miRNAs Drive Cross-Task Improvements in Drug Mechanistic Modeling —
• Hybrid GPU-CPU Retrieval for Personalized Search at Ultra-Large Scale —
• Authorization Revocation for Long-Running AI Agents: Root-Scoped Quiescence under Delegation and Asynchronous Execution —
• Multi-Subject Pretraining Enables Short-Calibration Personalization for Closed-Corpus Surface EMG Speech Decoding —
• GameASG-Bench: Benchmarking Autonomous Software Generation for Game Development —
• FairLMs: A Turnkey Library for Fairness in Language Models —
• Identifying Security Platform Product Abuse with Machine Learning —
• Fast And Accurate Text Content File Type Identification —
• An Introduction to Compression-Based Machine Learning —
• Diagonalized Attention for Individualized Regression: Latent-Row Localization and Prediction —
• Sparse Identification for Automatic Large-Scale Screening: A Constraint-Aware Framework with Ultra Fast Decoding Algorithm —
• Why most neutron star low-mass X-ray binaries accrete transiently: an evolutionary study of transient and persistent phases —
• LEGIT: Credentialing Protocol for Trustworthy AI Agent Marketplaces —
• Deep Reinforcement Learning with Buffered Quantile Objectives —
• Routine Blood Tests Outperform CRP for Distinguishing Bacterial From Viral Infection in Children —
• Patchy Reionization Delays Quenching in Isolated Ultra-faint Dwarf Galaxies —
• Co-Evolving Zero-Day Jamming: Adaptive Attack Synthesis and Graph Attention-Based Online Detection —
• A large overdensity of LRD-like galaxies discovered by JWST in the SPT2349-56 protocluster at z=4.30 —
• Conformal Privacy Auditing: Calibrated Re-identification Attacks with Statistical Guarantees —
• Robust Dual-Regularized Variable Selection under Outlier Contamination —
• A Gamma-ray Emitting Head-Tail Radio Galaxy Swimming Through the Hot Medium of the Merging Cluster Abell 3627 —
• CESBench: Benchmarking Large Language Models on Cryptographic Engineering Security for IoT Devices —
• IntBMoE: Integrating Block-Level Conditioning into Expert Composition for Full-Participation Mixture-of-Experts —
• From Memory to Behavior: A Behavior-Aware Role-Playing Framework for Social Media Influencers —
• From Observation to Ground Truth: Reconstruction Losses of Galaxy Imaging Under Heteroscedastic Noise —
• Beyond Atomic Tokens: Factorizing Syllables for Language Model Pretraining —
• Hiding in Plain Sight: A Diffusion-based Mitigation of Geolocation Privacy Leakage in Vision-Language Models —
• Batched Paillier-Based Hamming-Distance Computation over Binary Embeddings —
• Fountain pattern of baryon cycle revealed in galaxy ecosystems —
• ArenaFlow: From Trajectory Ranking to Hierarchical Credit Propagation for Open-Ended Agent RL —
• Knowledge-Graph-Augmented Chronos-2 for HEC-RAS Surrogate Forecasting —
• Probabilistic Forecasting of Business Process Executions with Neural Temporal Point Processes —
• Prediction Dynamics in Depth-Recurrent Language Models —
• AgentVidBench: A Multi-Hop Video Question Answering Benchmark for Evaluating MLLM Agents —
• Consistent Relexicalization of Clinical Documents using Graph-Based Approach —
• Offline Multimodal Large Language Models for Decision Support in Air Operations —
• WS-NeRF: A Mamba-Driven World-State-Aware Adaptive Deblurring Neural Radiance Field —
• Omni Demand Understanding: A Benchmark for Contextual User-Intent Inference in Multimodal Interaction —
• Talking Past the Machine: Morality, Politeness, and Alignment in Human-AI Dialogue —
• Gamma-ray Quasi-periodic Oscillations in the Changing Look Blazar B2 1420+326 (OQ 334) —
• TrustBOM: A Scalable Architecture for Confidentiality-Preserving SBOMs Across Organizations —
• Brownian Heads for Deep ReLU Representations: Activation Mass and the Cost of Same-Sample Selection —
• DENSE: Distilling Agent Trajectories into Evidence-Grounded Shortcut Trees for Self-Refinement —
• Tracing the Evidence Behind Zero-Shot Time-Series Forecasting: A Source-First Taxonomy and Audit Framework —
• DEFEAT: Stitching Fragmented File I/O Contexts for Early Ransomware Detection —
• Decision-Focused Learning for Mean-Variance Portfolio Optimization via KKT-Based Reformulation —
• GVPO++: Group Variance Policy Optimization for LLM Post-Training and On-Policy Distillation —
• Think Locally, Refine Globally for Memory-Efficient 3D Reconstruction —
• Interference-Driven Clustered Optimisation for FM Spectrum Coordination —
• The high-velocity dark matter halo of the Milky Way in light of the LZ 248 keV event —
• Optimal Randomized Proper Online Learning —
• FootQuery: Future-Touchdown-Guided Retrieval from Depth History for Perceptive Humanoid Locomotion —
• Understanding LLM Quantization through Activation-Guided Compensation and Orthogonal Residuals —
• Improving the Predictive Performance of Bootstrap Aggregating by Dirichlet Resampling —
• Efficient Architecture Search under Leave-One-Subject-Out Evaluation —
• AtomEgo: Exploring Ego-Robot Integration for Embodied Foundation Model Pretraining —
• OmniVChat: Synthesizing, Benchmarking, and Training for Native Audio-Visual Dialogue —
• Risk-Aware Occupancy for Safety-Oriented End-to-End Autonomous Driving —
• 21-cm Constraints on S-wave Dark Matter Annihilation with Subhalo-enhanced Effects —
• Adaptive Rollout Truncation Based on Epistemic Uncertainty for Efficient Offline World Model Training —
• HE-Guardrail: A Homomorphic Guardrail Against Jailbreak Attacks for Encrypted Large Language Model Inference —
• Driving on Registers, Reasoning on Risk: Risk-Aware Occupancy for Register-Based End-to-End Autonomous Driving —
• An Orbital Interpretation of Decreasing Galaxy Velocity Distance Relations in Systems Dominated by Giant Elliptical Galaxies —
• Benchmarking Gender Bias in Machine Translation Evaluation Metrics across Occupations —
• LogicTrack: Auditing Reasoning Trajectories of Large Language Models with Formal Logic Solvers —
• PolyBridgeBench: Benchmarking Multimodal LLMs for Physics-Grounded Bridge Design —
• The independence of the mid-infrared RR Lyrae Period-Luminosity relation from metallicity from a study of three globular clusters in the Large Magellanic Cloud —
• The Communication Bottleneck: A Round-Trip Study of Tree-Structured Expression Serialization in Language Models —
• 2nd Place Solution to the HANDS 2026 Workshop Challenge-Dexterous Grasp Motion Track: Single-Shot Trajectory Warping for Grasp Motion Generation —
• ServeGuard: Verifiable, Bounded-Residual Confinement of Operator-Invisible Channels Without Revealing the Certified Read Factor —
• Learning-to-Optimize as the Missing Architectural Layer of AI-Native Networks —
• VidOmni-Bench: A Benchmark for Fine-Grained Video Understanding via Spatio-Temporal Event Verification across Complexity and Duration —
• What Must Survive? Exact Task-Information--State Frontiers for Resource-Sufficient Learning —
• IncentRL: The Trade-Off Between Preference Guidance and Task Performance —
• OpenMAS-GCom. A Diagnostic Benchmark for Graph-enhanced Multi-Agent Systems —
• Critical sets of Latin squares based on autoparatopisms —
• MACE: Memory-Agent Co-Evolution with Adaptive Memory Graphs for Multi-Agent Systems —
• Purification and Regulation: Comorbidity-Aware Multi-Label Few-Shot Learning for Medical Image Classification —
• Dual-Interest Sequential Product Recommendation With Multi-Granular SSM —
• OneBid: A Unified Auto-Bidding Foundation Model for Diverse oCPX Advertising Scenarios —
• MIRAGE: Multi-Perspective Creative Language Model Reasoning with Reinforcement Learning Guidance —
• On Repulsive and Attractive Teachers: Separating Correctness from Behavior in Self-Distillation —
• GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions —
• Weighted Quantum Signal Processing: Low-Depth Polynomial Approximation with Applications to Kolmogorov-Arnold Networks —
• Et Tu, MacBook? Unprivileged Keystroke Inference and Context Profiling via the Built-in IMU Side Channel —
• CityLearn v3: A Configurable Simulation and Evaluation Framework for Realistic Control Studies of Renewable Energy Communities —
• Micro-Collaborative Poisoning: A Distributed Attack on RAG Systems —
• A note on oscillons: non-negligible metric fluctuations during preheating —
• Predictive Suppression Layers for Communication-Efficient Spiking Neural Networks —
• Periodic Neural Mapping for Unsteady Rotor-Blade Pressure and Aeroelastic Load Prediction —
• Evaluating In-Context Learning and Retrieval Strategies for Devanagari Post-OCR Correction —
• Hierarchical Hub Formation in Mon R2: Evidence for Cloud-to-core Gas Flows —
• Beyond Accuracy: Centroid-Guided Contrastive Loss for Structured Fraudulent Job Posting Detection —
• Reducing Barriers to Academic Support: Evaluating a Course-Specific RAG System for Addressing Help-Seeking Disparities in Higher Education —
• ALMA Chemical Evolution (ACE) survey: an overview -- extending dust and gas inference to low metallicities at cosmic noon —
• Trading Depth for Time in Recurrent Transformers —
• Combining astrometry with pulsar timing: the first joint analysis of very low frequency gravitational waves —
• Potential-Field Action Representation for Reinforcement Learning in Contact-Rich Manipulation —
• Calibrating Teacher--Student Discrepancy for On-Policy Distillation —
• One Prompt Does Not Fit All: Self-Meta-Evolve for Personalized Information Extraction —
• Detection is solved, delineation is not: what governs tooth segmentation on panoramic radiographs —
• A method for enhanced gamma-proton discrimination with imaging atmospheric Cherenkov telescopes —
• Steering LLMs Responses Towards Moral Foundations on the Norwegian MFQ-30 —
• Chinese Competitive Debating Dataset and Benchmark —
• Riemannian Neural Hamiltonian Flows: Geodesic Symplectic Transport and Interpretability —
• SynthDemo-RL: Breaking the Zero-Reward Barrier in VLA Adaptation with LLM-Guided Synthetic Demonstrations —
• Configurable Multi-Stage Vision Pipeline for Crop Disease and Pest Diagnosis —
• Wave-particle acceleration during supernova explosions —
• Analysing the Linearity of Linguistic Relations in Language Model Embedding Spaces —
• Beyond Gaussian Worlds: Latent Geometry Matters for JEPAs —
• KiDS J1447-0149: The first spatially resolved spectroscopy of a relic galaxy beyond the local universe. An old high-dispersion core embedded in a compact rotating stellar structure —
• Outcome-Conditioned End-Effector Geometry Across Vision-Language-Action Policies —
• Identification of gravitational lenses obscured by foreground light in the KiDS dataset using U-Nets and ResNets —
• LenNet: Direct Detection and Localization of Strong Gravitational Lenses in Wide-Field Sky Survey Images —
• When Steering Fails in Latent Reasoning: A Latent-to-Language Transition Gap —
• Rethinking Human-Aligned Evaluation: An Analysis of Semantic Metrics Beyond WER —
• Multi-Domain Clustering via Measure Quantization —
• Samsone: A Family of Open Small Audio Language Models for On-Device Inference —
• From Code Archival to Knowledge Graph: Bridging Software Heritage, COAR Notify and Wikidata —
• Accelerating Dense LLMs via L0-regularized Mixture-of-Experts —
• PRISM-BN: A Controlled Corpus and Benchmark for Text-to-Parameterized Bayesian Network Extraction —
• The Spoken Wikipedia Presentation Corpus —
• GUARD: Natural Forgetting in Large Reasoning Models via Guided Answer-Reasoning Distillation —
• Listen Before You Speak: Response Planning from Listener Facial Reactions for Conversational Speech Generation —
• CIPL: A Channel-Aware Framework for Recoverable Privacy Leakage in LLM Agents —
• Optimization Geometry of Equivalent Brownian RKHS Representations —
• Bayesian classification of astronomical spectra with class uncertainties —
• Lens Modeling and Cosmological Inference from an Impure Sample of Galaxy-Galaxy Strong Lenses —
• SpecQuant: Speculative Decoding with Multi-Parent Quantization for Adaptive LLM Inference —
• Morphology classification for galaxies in the Kilo Degree Survey using a label-efficient self-supervised learning framework —
• Faithful Neural Embeddings for 3D Exoplanet Climate Modeling —
• COBALT-BLUE: a fast and accurate model for gravitational self-lensing binary systems —
• X-ray polarization of the Compton-thin Seyfert galaxy NGC 5506 with IXPE —
• TERMon: Detecting Persistent Behavioral Threats in Edge AI via Hardware-Native Ternary Runtime Monitor —
• Some challenges for the long-term survival of Naiad, Neptune's innermost moon —
• A Framework to Quantify the Probability of Future Cyber Loss Events —
• CIBuzzBench: A Benchmark for Cross-Lingual Understanding of Chinese Internet Buzzwords —
• Verifiable Computation with Trusted Execution Environments and On-Chain Digital Rights Tokens —
• Beyond Disk Truncation: X-ray Reverberation Signatures of an Outflowing Corona —
• GEM-MPC: Balancing Exploration and Exploitation through Expert-Guided Planning —
• Core-collapse supernovae 2022prr, 2023ucy, and 2024ljc in the luminous infrared galaxy NGC 6745 —
• Balanced Prompt Adaptation against Entropy-Induced Collapse for Test-Time Binary Segmentation —
• Single-Loop Stochastic Projected Damped Extragradient Methods for Stochastic Nonconvex--(Strongly) Concave Minimax Optimization —
• World Modeling in Transformers —
• GraphSkillEvo: Evolutionary Optimization of Graph-Structured Agent Skills —
• ForceTwin: Physics-informed Digital Twins for Robotic Manipulation from Instrumented Human Interaction —
• ECG Mirage: Revealing and Mitigating the Underutilisation of ECGs in Vision-Language Models for Clinical Prediction —
• Bilevel Optimization of Topology and Hyperparameters (BOTH) —
• Complete Neural Electronic Initialization Accelerates Materials DFT —
• Neural composite likelihood estimation: simulation based inference for time series —
• Beyond Benchmark Scores: Auditing Medical Vision-Language Models for Chest X-Ray Tuberculosis Screening —
• Stability Analysis of the Proton Hammerhead Distribution Observed by Parker Solar Probe: Linear Theory and Fully Kinetic Simulations under Idealised Conditions —
• High Resolution Grid-based Simulations of the Warm-Hot Intergalactic Medium —
• From Pretraining to Proficiency: Real-World Subtask RL for Long-Horizon Manipulation with Minimal Human Intervention —
• Per-Aetiology Contrastive Severity Embeddings with Phonological Pseudo-Labelling for Multilingual Dysarthric Speech —
• RegKT: Interpretable and Robust Deep Knowledge Tracing With IRT-Regularizer —
• CASCADE Against Jailbreaks: Combination Across Stages with Controlled Attack-Defense Evaluation —
• LLM-Generated Feature Pools for Time Series Anomaly Detection —
• An Agentic Just-in-Time Adaptive Intervention System for Personalized Sleep Support: Proof-of-Concept Study with N of 1 Data —
• FLUMEN: Neural Emulator of an Advanced Stochastic Weak Lensing Model —
• JWST MEP - A World Under Spotty Starlight: Detection of CO2 and H2O in the Hot Saturn WASP-52b with JWST NIRSpec G395H —
• MIST: Multimodal Survival Prediction with Genomic-Guided Histology Attention —
• How Many Posterior Samples? Calibrated Stopping for Adaptive Sensing —
• Matrix AdaGrad: Row-wise and Column-wise Adaptive Subgradient Methods —
• Adaptive Uncertainty-Aware Modeling and Stochastic Radial Basis Function Predictive Control for Personalized Fluid Resuscitation —
• RheoSampling: Resolving the One-Hot Dilemma in Stochastic Dynamic-Tree Speculative Decoding —
• Touvigation: Embodied Adaptive Object Acquisition for Blind and Low-Vision Users in Unfamiliar Indoor Environments —
• Federated Deep Clustering Networks for High-Dimensional and Heterogeneous Data —
• SN 2026dix: a nearby, transitional Type IIb/Ib supernova consistent with a warm supergiant progenitor —
• Intrinsic Tidal Shear at the Largest Scales: Galaxy Multiplet Alignment in DESI DR2 —
• EnterpriseVal: Quantifying the Efficacy, Reliability and Value of Generative AI in the Enterprise —
• Reusing Latent Speech Representations for Query-Conditioned Topic Localization in Transcripts —
• The Weight Is Over - Interactive Diffusion on Consumer GPUs —
• Orbital and stellar parameter determination of eclipsing beta Cep pulsators —
• Supernova origin of galactic turbulence revealed by superbubbles —
• GIARPS simultaneous optical and infrared high-spectral-resolution observations of Cepheids I. Limb darkening from cross-correlated radial velocities —
• Do Personality-Tuned LLMs Make Better Social Agents? —
• Watermarkable Multi-Draft Speculative Sampling via Poisson Processes —
• TrialAtlas: Multi-Agent Research Organization for Clinical Trial Design and Optimization —
• A two-component profile of baryonic feedback: hydrostatic and diffuse gas —
• AutoRecLab: Describe the Experiment, Get the Code! —
• Neural Cellular Automata Learn General Features in their Hidden Channels —
• Chronosphere: Space-Time Tessellation of Local Climate Experts —
• SFPF: Spatio-Frequency Polarization Fingerprint for Anomalous Wireless Device Detection —
• Geometric Mean Pooling for Equal-Weight Multiplicative Coarse-Graining —
• Benchmarking the Explanatory Quality of Open-Weight Vision-Language Models in Face Recognition —
• Near-Optimal Acceleration for Smooth p / q Nondual Convex First-Order Oracle Optimization —
• Time-resolved JWST Retrieval Analysis of the Coldest Brown Dwarf: Temperature and Chemical Variations in WISE 0855-07 —
• Detecting Pretraining Data in Large Language Models from a Free-Energy Perspective —
• Structure and astrophysical properties of dark energy admixed neutron stars —
• LLMs as Feature Engineers for Text-and-Tabular Prediction —
• ExpBoN: Exponential-Noise Best-of- n for Efficient Test-Time LLM Alignment —
• Intervention Granularity Matters: Coherent Treatment Bundles in Counterfactual Simulation with Clinical World Models —
• Beyond Kinematics: Benchmarking Simulation Fidelity for Muscle-Driven Imitation Learning —
• Riemannian Simultaneous Inference for Tangent Vector Field Regression —
• Modified Gravity from growth data: goodness-of-fit and gravitational coupling —
• Nonlinearly Causal General-Relativistic Two-Fluid Dissipative Magnetohydrodynamics —
• What Should We Ask Next? Retrieval-Aware Question Learning for Interactive ReID —
• Kinks vs. Smoothness: Identifiability of Real Analytic nICA for Laplace-like Sources —
• Stellar ages from [C/N] in giant stars: applicability and limitations —
• Euclid: Quick Data Release (Q1) -- Optical hotspots in powerful radio galaxies —
• Joint Remaining Useful Life Prediction and Capacity Estimation of Lithium-Ion Batteries Using Partial-Charging Data —
• Stellar Yields of Rapidly Rotating Population III Stars for the High Redshift Universe —
• AutoViewMem: Self-Configuring Orthogonal Views for Conversational Long-Term Memory —
• End-to-End Hard-Label Cryptanalytic Model Extraction Using Efficient Sign Recovery —
• When Should a Failing Robot Ask? Initiating Corrective Human-Robot Dialogue from Audited Sensor Evidence —
• Guiding Agents of Quantum Games to Equilibrium using Matrix Exponential Fixed-Point Iteration —
• Learning to Move Cities: Deep Meta-Models and Reinforcement Policies for Calibration and Control in Urban Networks —
• RACER: Role-Aligned Competence Estimation for Human-AI Routing —
• Provisional Reachability: Containing Agents by Making Every Crossing Revocable —
• Schedule optimization for tau-leaping in masked discrete diffusion —
• Learning Cardiac Features: ECG Biometrics Across Time and Exercise —
• NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilities —
• Multiplicative Optimism for Constant Regret in Games —
• The JWST Ultramassive Galaxy Sample -- I. Overview, Photometry, and Stellar Kinematics: Selecting Optimal Template Subsets via Non-Negative Matrix Factorization —
• The JWST Ultramassive Galaxy Sample -- II. Supermassive Black Hole Masses for 8 Extreme Early-Type Galaxies via Jeans Anisotropic Modelling of High-Resolution Integral-Field NIRSpec Stellar Kinematics —
• Stochastic black hole growth tracks stellar mass in ultramassive galaxies —
• Time series generation with spectrally aligned latent flow matching —
• Moral Entropy: Auditing Bias and Uncertainty in Moral Judgment —
• Assessment of Machine Learning-Based Critical Heat Flux Models in the CTF Subchannel Code for Square Rod Bundle Prediction —
• A Lie Detector Test for Language Models: Reading Knowledge a Model Won't Reveal —
• Bayesian Belief Layer for Controllable Opinion Dynamics in LLM Agents —
• RecreationWorld: Scalable and Verifiable Environments for Hybrid Computer-Use Agents —
• Abstention and Noise Filtering: Two Missing Primitives of Softmax Attention —
• DiaVLo: Diagnosing Behaviours of Vision-Language Models —
• COMPLEX: A Closed-Form Certified Embedding of Multiparameter Persistence Modules —
• The Supersingular Isogeny Problem in Time and Memory p 1/3+o(1), Unconditionally —
• Comparative analysis of Neural Networks approaches for Initial Orbit Determination of Near-Earth Objects —
• Multiband Color Monitoring of 3I/ATLAS through Ground-Based Relay Observations —
• Rotational Mapping and Regional Atmospheric Retrievals of Variable Brown Dwarfs: Application to Luhman 16B and SIMP 0136 —
• The optical high-resolution spectral archive of the HERMES spectrograph mounted on the 1.2m Mercator telescope —
• Searching for Massive Remnant Planetary System Hosts with White Dwarf Gaia XP Spectra —
• QuranicMMLU: A Cognitively-Aware Benchmark for Evaluating Generative AI Solutions on Quranic Linguistic Knowledge —
• Gricea: An Open Science Platform for Conversational AI Research —
• lambda-Controlled GRPO: Turning Flow-Matching Ratio Instability into a Budgeted Resource —
• Asteroids Impacting the Solar System Planets and the Moon. I. Collision Rates from N-body Simulations —
• An Interpretable Memory Decision Controller for LLM Agents Based on Three-Signal Complementarity: Decoupling Confidence and Consistency —
• Asteroids Impacting the Solar System Planets and the Moon. II: Comparison with Observational Impact Records —
• Asteroids Impacting the Solar System Planets and the Moon. III. Real Impacts from Known Objects and LSST Discovery Predictions —
• Available Guardrails: Certifying Selective Prediction across ML Systems —
• Particle Competition and Cooperation for Robust Graph Convolutional Network Learning Under Label Noise —
• Benchmarking World Models for Continual Learning on Compositional Tasks —
• Predictable Failure in Multi-Hop Retrieval: Score-Distributional Confidence Scoring and Abstention —
• Non-linear halo bias for accurate modelling of cosmic infrared background anisotropies —
• Resurrecting Electroweak Dark Matter via Type-II Seesaw in light of recent LZ Event —
• BrainWideBench: Benchmarking large-scale pretraining and across-animal transfer in multi-region neural recordings —
• Value-Sensitive Delegation in Everyday AI Agent Use: Evidence from OpenClaw —
• CodeMidas: Scaling Agentic Coding RL Environments from Code Itself —
• Discovery, Characterization, and Potential Origins of a Stream in the Stellar Halo of Nearby LMC-Mass Galaxy NGC 55 —
• Multi-Epoch Stability in the Rotational Modulation of the Planetary-Mass Companion Ross 458C —
• The AGN Channel in 3D: Scattering Belts and the Importance of Eccentricity in the Black Hole Population —
• APort Vault: Benchmarking AI Agent Payment Authorization with the Open Agent Passport —
• Structuring occupational accident narratives for cross-sector safety analysis: Transferability of accident-process role classification —
• On the X-ray emission from massive star clusters and their evolving superbubbles II. Detailed analytics and observational effects —
• Designer-RSI: Evolving Procedural Memory from User Traffic for Agentic Graphic Design —
• Gravitational and mass distribution effects on stationary superwinds —
• An analysis of the turbulence in the central region of M 42 through structure functions —
• Optimal Scoring Rule Design under Partial Knowledge —
• PI -- Multimodal Planetary Defense —
• Modeling general-relativistic plasmas with collisionless moments and dissipative two-fluid magnetohydrodynamics —

Important terms

Nuclear Pasta

Exotic structures found in the inner crust of neutron stars. These include shapes like connected rods or slices with holes, which researchers are now modeling using new density functional theory methods without needing manual geometric assumptions.

Activation Steering

A technique used to manipulate how large language models behave by boosting or suppressing their internal confidence signals. This helps researchers prove that models use these signals to decide when to abstain from answering a question.

Recursive Self-Improvement

The theoretical idea of programs that can introspect and evolve themselves. While some mathematical theorems suggest this is possible, current AI architectures face structural bottlenecks that prevent them from reaching the necessary level of self-access.

Agentic Communication

A method where multiple AI agents talk to each other to solve problems. This collaborative approach allows teams of agents to outperform single models and even solve complex tasks that are impossible for an individual agent alone.

GEM-MPC

A new framework for robotics and continuous control that balances following a plan with exploring new actions. It uses special distillation techniques to learn from old data efficiently without wasting computational power on outdated information.