AI Agents

[D] Your Agent, Their Asset: Real-world safety evaluation of OpenClaw agents (CIK poisoning raises attack success to ~64–74%)

Paper: https://arxiv.org/abs/2604.04759 This paper presents a real-world safety evaluation of OpenClaw, a personal AI agent with access t...

New privacy tool helps detect when AI agents become double agents

RIT cybersecurity researchers have developed AudAgent, a tool that detects when agentic AI collects, processes, or shares highly sensitiv...

Microsoft’s GitHub Sees Booming Traffic—and Outages—as AI Agents Flood Platform

[2511.10515] Mastering Olympiad-Level Physics with Artificial Intelligence

This paper presents LOCA, an AI framework designed to tackle Olympiad-level physics problems by breaking down complex reasoning into mana...

[2510.20091] CreativityPrism: A Holistic Evaluation Framework for Large Language Model Creativity

The paper presents CreativityPrism, a comprehensive framework for evaluating the creativity of large language models (LLMs) across variou...

[2510.12768] Uncertainty Matters in Dynamic Gaussian Splatting for Monocular 4D Reconstruction

This paper presents USplat4D, a novel framework for monocular 4D reconstruction that incorporates uncertainty in dynamic Gaussian splatti...

[2602.10956] Stochastic Parroting in Temporal Attention -- Regulating the Diagonal Sink

The paper explores the challenges of spatio-temporal models in machine learning, focusing on biases in temporal attention mechanisms and ...

[2510.06200] StarEmbed: Benchmarking Time Series Foundation Models on Astronomical Observations of Variable Stars

The paper introduces StarEmbed, a benchmark for evaluating time series foundation models on astronomical observations of variable stars, ...

[2510.04694] Multilingual Routing in Mixture-of-Experts

This paper explores multilingual routing in Mixture-of-Experts (MoE) architectures, revealing how these models handle multilingual data a...

[2602.05139] Adaptive Exploration for Latent-State Bandits

The paper presents adaptive exploration strategies for latent-state bandits, addressing challenges in reward estimation and action select...

[2509.05249] COGITAO: A Visual Reasoning Framework To Study Compositionality & Generalization

COGITAO introduces a novel framework for studying compositionality and generalization in visual reasoning, offering extensive task genera...

[2508.08177] MedReasoner: Reinforcement Learning Drives Reasoning Grounding from Clinical Thought to Pixel-Level Precision

The paper introduces MedReasoner, a framework that utilizes reinforcement learning for precise medical reasoning and pixel-level groundin...

[2601.11616] Mixture-of-Experts as Soft Clustering: A Dual Jacobian-PCA Spectral Geometry Perspective

This paper explores Mixture-of-Experts (MoE) architectures through a geometric lens, analyzing their impact on function representation an...

[2508.01067] Expressive Power of Graph Transformers via Logic

This paper explores the expressive power of graph transformers, comparing their capabilities under different logical frameworks, particul...

[2506.08822] FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistency

The paper presents FreqPolicy, a novel flow-based visuomotor policy that enhances efficiency in robotic manipulation by imposing frequenc...

[2511.22581] High entropy leads to symmetry equivariant policies in Dec-POMDPs

This paper explores how high entropy regularization in Dec-POMDPs leads to symmetry equivariant policies, ensuring convergence to a consi...

[2505.03795] Modeling Human Behavior in a Strategic Network Game with Complex Group Dynamics

This article explores modeling human behavior in strategic network games, focusing on the Junior High Game (JHG) and comparing various be...

[2511.03710] Shrinking the Variance: Shrinkage Baselines for Reinforcement Learning with Verifiable Rewards

This article presents a novel approach to reducing variance in reinforcement learning through shrinkage baselines, enhancing training sta...

[2504.08603] FindAnything: Open-Vocabulary and Object-Centric Mapping for Robot Exploration in Any Environment

The paper presents FindAnything, a framework for open-vocabulary and object-centric mapping that enhances robot exploration in unknown en...

[2510.24318] Transformers can do Bayesian Clustering

The paper presents Cluster-PFN, a Transformer-based model for unsupervised Bayesian clustering, demonstrating improved accuracy and speed...

[2510.19753] Transformers Provably Learn Algorithmic Solutions for Graph Connectivity, But Only with the Right Data

The paper explores how Transformers can learn algorithmic solutions for graph connectivity, demonstrating that success depends on the tra...

[2503.12286] Integrating Chain-of-Thought and Retrieval Augmented Generation Enhances Rare Disease Diagnosis from Clinical Notes

This article presents a novel approach combining Chain-of-Thought (CoT) and Retrieval Augmented Generation (RAG) to improve rare disease ...

[2502.17863] A Survey: Spatiotemporal Consistency in Video Generation

This survey reviews advancements in spatiotemporal consistency in video generation, addressing challenges and methodologies in creating c...