Daily Arxiv

전 세계에서 발간되는 인공지능 관련 논문을 정리하는 페이지 입니다.
본 페이지는 Google Gemini를 활용해 요약 정리하며, 비영리로 운영 됩니다.
논문에 대한 저작권은 저자 및 해당 기관에 있으며, 요약본 공유 시 출처만 명기하면 됩니다.
This service is supported by Google Gemini.

MaxShapley: Towards Incentive-compatible Generative Search with Fair Context Attribution

NORi: An ML-Augmented Ocean Boundary Layer Parameterization

Open-Set Domain Adaptation Under Background Distribution Shift: Challenges and A Provably Efficient Solution

HarmonicAttack: An Adaptive Cross-Domain Audio Watermark Removal

Data-driven Acceleration of MPC with Guarantees

Adaptive Residual-Update Steering for Low-Overhead Hallucination Mitigation in Large Vision Language Models

PlantTraitNet: An Uncertainty-Aware Multimodal Framework for Global-Scale Plant Trait Inference from Citizen Science Data

Noise-corrected GRPO: From Noisy Rewards to Unbiased Gradients

Needles in the Landscape: Semi-Supervised Pseudolabeling for Archaeological Site Discovery under Label Scarcity

Dr.LLM: Dynamic Layer Routing in LLMs

WARC-Bench: Web Archive Based Benchmark for GUI Subtask Executions

Proximal Diffusion Neural Sampler

Beyond Majority Voting: LLM Aggregation by Leveraging Higher-Order Information

HiDe: Rethinking The Zoom-IN method in High Resolution MLLMs via Hierarchical Decoupling

Jailbreaking on Text-to-Video Models via Scene Splitting Strategy

YAC: Bridging Natural Language and Interactive Visual Exploration with Generative AI for Biomedical Data Discovery

ORCA: An Agentic Reasoning Framework for Hallucination and Adversarial Robustness in Vision-Language Models

Digital Voices of Survival: From Social Media Disclosures to Support Provisions for Domestic Violence Victims

PiKV: KV Cache Management System for Mixture of Experts

CoLD: Counterfactually-Guided Length Debiasing for Process Reward Models in Mathematical Reasoning

Fine-tuning Large Language Model for Automated Algorithm Design

Mechanistic Interpretability Needs Philosophy

Recursive Entropic Risk Optimization in Discounted MDPs: Sample Complexity Bounds with a Generative Model

Spatial-MLLM: Boosting MLLM Capabilities in Visual-based Spatial Intelligence

Quantifying the Generalization Gap in Seizure Detection: A Large-Scale Empirical Benchmark via the SzCORE Challenge

Slot-MLLM: Object-Centric Visual Tokenization for Multimodal LLM

GraphPINE: Graph Importance Propagation for Interpretable Drug Response Prediction

Adaptive Camera Sensor for Vision Models

Faster-GCG: Efficient Discrete Optimization Jailbreak Attacks against Aligned Large Language Models

Monitoring AI-Modified Content at Scale: A Case Study on the Impact of ChatGPT on AI Conference Peer Reviews

Deep Neural Network for Musical Instrument Recognition using MFCCs

Latent Action Reparameterization for Efficient Agent Inference

Memory-Guided Tree Search with Cross-Branch Knowledge Transfer for LLM Solver Synthesis

The Capability Paradox: How Smarter Auditors Make Multi-Agent Systems Less Secure

CBT-Audio: Evaluating Audio Language Models for Patient-Side Distress Intensity Estimation in CBT Session Recordings

Recall Isn't Enough: Bounding Commitments in Personalized Language Systems

Skim: Speculative Execution for Fast and Efficient Web Agents

Learning Bilevel Policies over Symbolic World Models for Long-Horizon Planning

ALSO: Adversarial Online Strategy Optimization for Social Agents

$\pi$-Bench: Evaluating Proactive Personal Assistant Agents in Long-Horizon Workflows

ChromaFlow: A Negative Ablation Study of Orchestration Overhead in Tool-Augmented Agent Evaluation

Network-Aware Bilinear Tokenization for Brain Functional Connectivity Representation Learning

VERA-MH: Validation of Ethical and Responsible AI in Mental Health

TMAS: Scaling Test-Time Compute via Multi-Agent Synergy

Active Testing of Large Language Models via Approximate Neyman Allocation

When to Re-Commit: Temporal Abstraction Discovery for Long-Horizon Vision-Language Reasoning

2.5-D Decomposition for LLM-Based Spatial Construction

Autogenesis: A Self-Evolving Agent Protocol

ReSS: Learning Reasoning Models for Tabular Data Prediction via Symbolic Scaffold

Contrastive Reasoning Alignment: Reinforcement Learning from Hidden Representations

A Hybrid Modeling Framework for Crop Prediction Tasks via Dynamic Parameter Calibration and Multi-Task Learning

Governing Evolving Memory in LLM Agents: Risks, Mechanisms, and the Stability and Safety Governed Memory (SSGM) Framework

The World Won't Stay Still: Programmable Evolution for Agent Benchmarks

Phase-Aware Mixture of Experts for Agentic Reinforcement Learning

Sonar-TS: Search-Then-Verify Natural Language Querying for Time Series Databases

TSR: Trajectory-Search Rollouts for Multi-Turn RL of LLM Agents

Nonlinearity as Rank: Generative Low-Rank Adapter with Radial Basis Functions

High-quality generation of dynamic game content via small language models: A proof of concept

Distributional AGI Safety

BuildArena: A Physics-Aligned Interactive Benchmark of LLMs for Engineering Construction

From Refusal to Recovery: A Control-Theoretic Approach to Generative AI Guardrails

ARM: Discovering Agentic Reasoning Modules for Generalizable Multi-Agent Systems

Learning Efficient Guardrails for Compliance

Extreme Self-Preference in Language Models

Artificial Phantasia: Emergent Mental Imagery in Large Language Models

IR-Agent: Expert-Inspired LLM Agents for Structure Elucidation from Infrared Spectra

Artificial Intelligence, conceptual metaphors and conceptual engineering: Are AI-based framings of human behaviour and cognition successful?

Entry-level guide to the use of large language models for medical research

Atoms of Thought: Universal EEG Representation Learning with Microstates

Long-term Power Grid Planning via Answer Set Programming

Rethinking Visual Attribution for Chest X-ray Reasoning in Large Vision Language Models

Less Back-and-Forth: A Comparative Study of Structured Prompting

Beyond Prediction Accuracy: Target-Space Recovery Profiles for Evaluating Model-Brain Alignment

Toto 2.0: Time Series Forecasting Enters the Scaling Era

k-Inductive Neural Barrier Certificates for Unknown Nonlinear Dynamics

Beyond Isotropy in JEPAs: Hamiltonian Geometry and Symplectic Prediction

Draft Less, Retrieve More: Hybrid Tree Construction for Speculative Decoding

INSHAPE: Instance-Level Shapelets for Interpretable Time-Series Classification

ThoughtTrace: Understanding User Thoughts in Real-World LLM Interactions

What Do Evolutionary Coding Agents Evolve?

BalanceRAG: Joint Risk Calibration for Cascaded Retrieval-Augmented Generation

VL-DPO: Vision-Language-Guided Finetuning for Preference-Aligned Autonomous Driving

Probability-Conserving Flow Guidance

CopT: Contrastive On-Policy Thinking with Continuous Spaces for General and Agentic Reasoning

Towards LLM-Assisted Architecture Recovery for Real-World ROS~2 Systems: An Agent-Based Multi-Level Approach to Hierarchical Structural Architecture Reconstruction

PromptRad: Knowledge-Enhanced Multi-Label Prompt-Tuning for Low-Resource Radiology Report Labeling

Does Code Cleanliness Affect Coding Agents? A Controlled Minimal-Pair Study

When Critics Disagree: Adaptive Reward Poisoning Attacks in RIS-Aided Wireless Control System

Training Neural Networks with Optimal Double-Bayesian Learning

LLM Benchmark Datasets Should Be Contamination-Resistant

A Case for Agentic Tuning: From Documentation to Action in PostgreSQL

Learning with Foresight: Enhancing Neural Routing Policy via Multi-Node Lookahead Prediction

Block-Sphere Vector Quantization

Detecting Fluent Optimization-Based Adversarial Prompts via Sequential Entropy Changes

World-Ego Modeling for Long-Horizon Evolution in Hybrid Embodied Tasks

GEM: GPU-Variability-Aware Expert to GPU Mapping for MoE Systems

A Measure-Theoretic Analysis of Reasoning: Structural Generalization and Approximation Limits

StruMPL: Multi-task Dense Regression under Disjoint Partial Supervision and MNAR Labels

Breaking Modality Heterogeneity in Low-Bit Quantization for Large Vision-Language Models

Real-Time Parallel Counterfactual Regret Minimization

TIME: A Multi-level Benchmark for Temporal Reasoning of LLMs in Real-World Scenarios

Created by

Haebom

저자

Shaohang Wei, Wei Li, Feifan Song, Wen Luo, Tianyi Zhuang, Haochen Tan, Zhijiang Guo, Houfeng Wang

개요

본 논문은 대규모 언어 모델(LLM)의 실세계 시간적 추론 능력 향상을 위해, 실세계의 도전 과제(집약적인 시간 정보, 빠르게 변화하는 사건 역학, 사회적 상호 작용의 복잡한 시간적 의존성)를 고려한 다층적 벤치마크 TIME을 제안합니다. TIME은 38,522개의 질의응답 쌍으로 구성되며, 3개의 수준과 11개의 세분화된 하위 작업으로 나뉩니다. TIME-Wiki, TIME-News, TIME-Dial 세 개의 하위 데이터셋은 서로 다른 실세계 과제를 반영합니다. 논문에서는 추론 모델과 비추론 모델에 대한 광범위한 실험 결과와 다양한 실제 시나리오 및 작업 전반의 시간적 추론 성능에 대한 심층 분석, 그리고 테스트 시간 스케일링의 영향을 요약합니다. 추가적으로, 미래 연구와 표준화된 평가를 촉진하기 위해, 사람이 주석을 단 하위 집합인 TIME-Lite도 공개합니다. 코드는 https://github.com/sylvain-wei/TIME 에서, 데이터셋은 https://huggingface.co/datasets/SylvainWei/TIME 에서 이용 가능합니다.