Search: “safety”

Path-Space Mirror Descent for On-Policy Reinforcement Learning under the Generalized Schrodinger Bridge

arXiv:2603.21621v2 Announce Type: replace Abstract: Classical on-policy algorithms such as PPO and mirror descent policy optimization provide stable proximal policy updates through tractable action li

Safety

PEARL: Training Socratic Tutors with Pedagogically Aligned Reinforcement Learning

DGX agent

arXiv:2605.29582v1 Announce Type: cross Abstract: Large Language Models (LLMs) have shown promise as educational tutors, yet effective tutoring requires more than solving problems: it must provide pro

safetyarxiv-cs-cl

29 May 2026

Permutation-Invariant Spectral Learning via Dyson Diffusion

arXiv:2510.08535v2 Announce Type: replace-cross Abstract: Diffusion models are central to generative modeling and have been adapted to graphs by diffusing adjacency matrix representations. The challen

PersonaAgent: Bridging Memory and Action for Personalized LLM Agents

arXiv:2506.06254v2 Announce Type: replace Abstract: Large Language Model (LLM) empowered agents have recently emerged as advanced paradigms that exhibit impressive capabilities in a wide range of doma

Phase-Conditioned Imitation Learning with Autonomous Failure Recovery for Robust Deformable Object Manipulation

arXiv:2605.29407v1 Announce Type: new Abstract: This paper presents a phase-conditioned, force-aware framework for robust deformable object manipulation. Standard imitation learning policies such as A

safetyarxiv-cs-ro

Plan, Don't Pose: Long Composite Motion Generation with Text-Aligned BFM

arXiv:2605.29906v1 Announce Type: new Abstract: Text-to-motion (T2M) generation has broad applications in character animation, virtual avatars, and human-robot interaction. Existing methods typically

Position: Stop Chasing the C-index when Evaluating Survival Analysis Models

arXiv:2506.02075v2 Announce Type: replace-cross Abstract: The current state of evaluation in survival analysis is plagued by the persistent use of evaluation metrics in ways that are misaligned with t

Practitioner Beliefs and Behaviors in AI-Enhanced Education: DOT Framework Survey Evidence

arXiv:2605.29041v1 Announce Type: new Abstract: This study reports findings from a cross-sectional survey (n = 72) of higher education practitioners examining beliefs, behaviors, and institutional con

PRO-CUA: Process-Reward Optimization for Computer Use Agents

arXiv:2605.29119v1 Announce Type: new Abstract: Computer use agents (CUAs) have shown strong potential for automating complex digital workflows, yet their training remains constrained by costly live e

Promoting Generalization for Exact Solvers via Adversarial Instance Augmentation

arXiv:2310.14161v2 Announce Type: replace Abstract: Machine learning has been successfully applied to improve the efficiency of Mixed-Integer Linear Programming (MILP) solvers. However, the learning-b

Provably Secure Agent Guardrail

arXiv:2605.29251v1 Announce Type: new Abstract: As large language models transition from bounded generative engines to agents with expansive execution privileges, AI going out of control precipitates

Quantifying and Optimizing Simplicity via Polynomial Representations

arXiv:2605.29823v1 Announce Type: new Abstract: Deep networks often exhibit a preference for 'simple' solutions, and such a simplicity bias is widely believed to play a key role in generalization. Yet

quick, name a prominent company that achieved a major win via tokenmaxxing!

quick, name a prominent company that achieved a major win via tokenmaxxing! I've yet to see, or hear of a company that is winning against its competition, because said company is spending more on AI t

Quotient DAGs for Off-Policy Evaluation:Forward-Flow Importance Sampling and Exact Slate Propensities

arXiv:2605.29500v1 Announce Type: cross Abstract: Off-policy evaluation estimates how a target policy would perform using data collected by a different behavior policy, which is crucial when online te

Real-Time Retargeting Using Controllability Boundary for Chandrayaan-3 Lunar Landing

arXiv:2605.29412v1 Announce Type: cross Abstract: This paper presents the real-time retargeting guidance policy developed for the Chandrayaan-3 lunar landing mission. The baseline guidance generates a

Reasoning While Asking: Transforming Reasoning Large Language Models from Passive Solvers to Proactive Inquirers

arXiv:2601.22139v2 Announce Type: replace-cross Abstract: Reasoning-oriented Large Language Models (LLMs) have achieved remarkable progress with Chain-of-Thought (CoT) prompting, yet they remain funda

ReasonLight: A Multimodal Foundation Model-Enhanced Reinforcement Learning Framework for Zero-Shot Traffic Signal Control

arXiv:2605.29425v1 Announce Type: new Abstract: Reinforcement learning (RL) has shown promise in traffic signal control (TSC). However, its reliance on predefined states limits responsiveness to obser

Recovering Policy-Induced Errors: Benchmarking and Trajectory Synthesis for Robust GUI Agents

arXiv:2605.29447v1 Announce Type: cross Abstract: While GUI agents have advanced rapidly, they often lack the robustness to recover from their own errors, hindering real-world deployment. To bridge th

safetyarxiv-cs-cl

Recurrent Structural Policy Gradient for Partially Observable Mean Field Games

arXiv:2602.20141v2 Announce Type: replace Abstract: Mean Field Games (MFGs) provide a principled framework for modelling interactions in large population systems. However, algorithmic progress has bee

Reinforcing Few-step Generators via Reward-Tilted Distribution Matching

arXiv:2605.26108v2 Announce Type: replace Abstract: Recent advances in few-step diffusion distillation have enabled efficient image generation, yet aligning these models with human preferences remains

Replicable Simulation-Based Robot Validation through Provenance

arXiv:2605.29973v1 Announce Type: new Abstract: Robot behavior is often validated through simulation-based testing, yet the replicability of such campaigns depends critically on transparent documentat

safetyarxiv-cs-ro

Representation Alignment Rests on Linear Structure

arXiv:2605.28870v1 Announce Type: cross Abstract: We investigate the Platonic Representation Hypothesis (PRH) through a tripartite statistical framework of representations: signal, bias, and noise. {1

Representation Signatures and Risk-Feedback Alignment in LLM Trading Agents

arXiv:2605.28850v1 Announce Type: new Abstract: We study behavioral alignment and representation dynamics of large language model (LLM) agents in financial decision environments. Using TradeArena, an

Resolution as a Direction: Vector-Panning Feature Alignment for Cross-Resolution Re-Identification

arXiv:2510.00936v2 Announce Type: replace Abstract: Cross-resolution person re-identification (CR-ReID) remains challenging in practical surveillance, where camera quality and capture distance lead to

Resolving Endpoint Underfitting in Diffusion Bridges via Noise Alignment

arXiv:2605.28962v1 Announce Type: new Abstract: Diffusion bridge models offer a powerful framework for connecting two data distributions, such as in image restoration and translation. Many existing me

REST3D: Reconstructing Physically Stable 3D Scenes from a Single Image

arXiv:2605.30338v1 Announce Type: new Abstract: Reconstructing physically stable 3D scenes from a single RGB image enables casual images to be converted into simulation-ready digital assets for applic

Review Arcade: On the Human Alignment and Gameability of LLM Reviews

arXiv:2605.28897v1 Announce Type: new Abstract: LLM-generated reviews for scientific papers are gaining considerable traction and are even being officially piloted by major conferences. We have to ass

RL2ML: Finite-Rollout Surrogate Objectives from Reinforcement Learning to Maximum Likelihood

arXiv:2605.30154v1 Announce Type: new Abstract: Correctness-based Reinforcement Learning with Verifiable Rewards (RLVR) trains language models from binary feedback on sampled outputs, but the objectiv

Rooted Absorbed Prefix Trajectory Balance with Submodular Replay for GFlowNet Training

arXiv:2603.00454v2 Announce Type: replace-cross Abstract: Generative Flow Networks (GFlowNets) enable fine-tuning large language models to approximate reward-proportional posteriors, but they remain p

RUBRIC-ARROW: Alternating Pointwise Rubric Reward Modeling for LLM Post-training in Non-verifiable Domains

arXiv:2605.29156v1 Announce Type: cross Abstract: Pointwise reward modeling offers critical signals for LLM post-training, yet struggles with absolute scoring in subjective, non-verifiable settings. R

safetyarxiv-cs-cl

Rubric-Guided Process Reward for Stepwise Model Routing

arXiv:2605.29310v1 Announce Type: new Abstract: Stepwise model routing improves the efficiency of Large Reasoning Models (LRMs) by assigning each reasoning step to a suitable model. Recent methods for

SAHG: Sector-Anisotropic Hyperbolic Graph Model for Social Bot Detection

arXiv:2605.30166v1 Announce Type: cross Abstract: LLM-driven social bots can generate fluent, human-like text, reducing the discriminative advantage of content-based detection alone. However, coordina

Same Evidence, Different Answers: Canonical-Context On-Policy Distillation for Multi-Turn Language Models

arXiv:2605.30251v1 Announce Type: cross Abstract: Large language models (LLMs) often solve a task when all instructions are given in a single prompt, but fail when the same information is revealed gra

Securing SIM-Assisted Wireless Networks via Quantum Reinforcement Learning

arXiv:2602.13238v2 Announce Type: replace-cross Abstract: Stacked intelligent metasurfaces (SIMs) have recently emerged as a powerful wave-domain technology that enables multi-stage manipulation of el

Self-Play Reinforcement Learning under Imperfect Information in Big 2

arXiv:2605.28863v1 Announce Type: cross Abstract: Imperfect-information multiplayer games test whether agents can act under hidden information, sparse rewards, and non-stationary opponents. We study t

SGMD: Score Gradient Matching Distillation for Few-Step Video Diffusion Distillation

arXiv:2605.30116v1 Announce Type: new Abstract: Distribution Matching Distillation (DMD) is a widely used paradigm for accelerating inference in few-step video diffusion models. However, DMD-style vid

Shame on you Apple, sending @mcuban’s thoughtful email to spam. Seriously? Glad I have learned not to trust your algorithm.

Gary Marcus criticizes Apple's email filtering system for incorrectly routing Mark Cuban's message to spam, highlighting a failure in Apple's spam detection algorithm. The post reflects concerns about

Statistical Embeddings for Similarity, Retrieval, and Interpretable Alignment of Numeric Tabular Datasets

arXiv:2605.30289v1 Announce Type: new Abstract: Numeric tabular datasets are the dominant data format in scientific practice, yet large language models lack native mechanisms for representing numeric

talking the party line; they switch, he switches. 🤣

This post by cognitive scientist Gary Marcus appears to comment humorously on political or ideological conformity, suggesting that people and leaders flip their positions based on party alignment rath

Teaching Values to Machines: Simulating Human-Like Behavior in LLMs

arXiv:2605.30036v1 Announce Type: new Abstract: Large Language Models (LLMs) demonstrate a remarkable capacity to adopt different personas and roles; however, it remains unclear whether they can manif

The Best of the Two Worlds: Harmonizing Semantic and Hash IDs for Sequential Recommendation

arXiv:2512.10388v2 Announce Type: replace-cross Abstract: Conventional Sequential Recommender Systems (SRS) typically assign unique hash IDs (HID) to construct item embeddings, which mainly capture co

The Confidence Shortcut: A Reasoning Failure Mode of Masked Diffusion Models

arXiv:2605.29123v1 Announce Type: new Abstract: Masked diffusion language models (MDMs) uniquely support any-order generation, with confidence-based decoding currently serving as the de facto standard

The Importance of Out-of-Band Metadata for Safe Autonomous Agents: The Redpanda Agentic Data Plane

arXiv:2605.29082v1 Announce Type: new Abstract: AI agents are increasingly expected to operate as digital employees: accessing enterprise data, making decisions, and taking actions autonomously. But a

The Sample Complexity of Multiclass and Sparse Contextual Bandits

arXiv:2605.29645v1 Announce Type: cross Abstract: We study contextual bandits in the stochastic i.i.d. setting, where a learner observes contexts drawn from an unknown distribution, selects actions fr

Theoretical Foundations and Effective Algorithms for Policy-Aware Simulator Learning

arXiv:2605.29032v1 Announce Type: new Abstract: Model-based reinforcement learning (MBRL) agents typically learn world models by minimizing predictive loss. However, powerful RL optimizers inevitably

Think Fast, Talk Smart: Partitioning Deterministic and Neural Computation for Structured Health Text Generation

arXiv:2605.29652v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly being used to generate health text from structured records such as wearable time series, biomarkers, vital

this is on track to be this year’s “what did Ilya see?” rumor that people love that has nothing to do with reality.

this is on track to be this year’s “what did Ilya see?” rumor that people love that has nothing to do with reality. Im calling BS on this story. 1. That would be 100,000 employees spending 5k/mo each

Today I learned that @Grimezsz has more courage in her pinky than Roon does in his entire cowardly body. Roon made excuses; Grimes defended …

Today I learned that @Grimezsz has more courage in her pinky than Roon does in his entire cowardly body. Roon made excuses; Grimes defended her views calmly and respectfully, like grownups should. Ope