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, 공유 시 출처만 명기하면 됩니다.

IG Parser: A Software Package for the Encoding of Institutional Statements using the Institutional Grammar

One-Step Offline Distillation of Diffusion-based Models via Koopman Modeling

J4R: Learning to Judge with Equivalent Initial State Group Relative Policy Optimization

Evaluating the efficacy of LLM Safety Solutions : The Palit Benchmark Dataset

From Assistants to Adversaries: Exploring the Security Risks of Mobile LLM Agents

Leveraging LLM Inconsistency to Boost Pass@k Performance

Sinusoidal Initialization, Time for a New Start

Enhancing Channel-Independent Time Series Forecasting via Cross-Variate Patch Embedding

Any-to-Any Learning in Computational Pathology via Triplet Multimodal Pretraining

Predicting Turn-Taking and Backchannel in Human-Machine Conversations Using Linguistic, Acoustic, and Visual Signals

ChromFound: Towards A Universal Foundation Model for Single-Cell Chromatin Accessibility Data

IP Leakage Attacks Targeting LLM-Based Multi-Agent Systems

RoboFAC: A Comprehensive Framework for Robotic Failure Analysis and Correction

LLM-DSE: Searching Accelerator Parameters with LLM Agents

Online Iterative Self-Alignment for Radiology Report Generation

Reachability Barrier Networks: Learning Hamilton-Jacobi Solutions for Smooth and Flexible Control Barrier Functions

Improving Medium Range Severe Weather Prediction through Transformer Post-processing of AI Weather Forecasts

BioCube: A Multimodal Dataset for Biodiversity Research

One Shot Dominance: Knowledge Poisoning Attack on Retrieval-Augmented Generation Systems

TCC-Bench: Benchmarking the Traditional Chinese Culture Understanding Capabilities of MLLMs

FALCON: False-Negative Aware Learning of Contrastive Negatives in Vision-Language Pretraining

GRoQ-Loco: Generalist and Robot-agnostic Quadruped Locomotion Control using Offline Datasets

Who You Are Matters: Bridging Topics and Social Roles via LLM-Enhanced Logical Recommendation

Artificial Intelligence Bias on English Language Learners in Automatic Scoring

Learning Long-Context Diffusion Policies via Past-Token Prediction

Fast Text-to-Audio Generation with Adversarial Post-Training

Unified Continuous Generative Models

Can LLM-based Financial Investing Strategies Outperform the Market in Long Run?

Technical Report: Quantifying and Analyzing the Generalization Power of a DNN

Prompting Large Language Models for Training-Free Non-Intrusive Load Monitoring

On-Device LLM for Context-Aware Wi-Fi Roaming

Efficient Fine-Tuning of Quantized Models via Adaptive Rank and Bitwidth

Understanding University Students' Use of Generative AI: The Roles of Demographics and Personality Traits

Adaptive Thinking via Mode Policy Optimization for Social Language Agents

LLM-hRIC: LLM-empowered Hierarchical RAN Intelligent Control for O-RAN

On the Boolean Network Theory of Datalog$^\neg$

Learning to Reason under Off-Policy Guidance

How Effective Can Dropout Be in Multiple Instance Learning ?

Learning Joint ID-Textual Representation for ID-Preserving Image Synthesis

Walk the Talk? Measuring the Faithfulness of Large Language Model Explanations

Cross-Document Cross-Lingual NLI via RST-Enhanced Graph Fusion and Interpretability Prediction

S1-Bench: A Simple Benchmark for Evaluating System 1 Thinking Capability of Large Reasoning Models

Beyond Self-Reports: Multi-Observer Agents for Personality Assessment in Large Language Models

Scaling Video-Language Models to 10K Frames via Hierarchical Differential Distillation

Scaling Test-Time Inference with Policy-Optimized, Dynamic Retrieval-Augmented Generation via KV Caching and Decoding

Breaking Language Barriers in Visual Language Models via Multilingual Textual Regularization

LogicQA: Logical Anomaly Detection with Vision Language Model Generated Questions

Conjuring Positive Pairs for Efficient Unification of Representation Learning and Image Synthesis

CRCE: Coreference-Retention Concept Erasure in Text-to-Image Diffusion Models

LED: LLM Enhanced Open-Vocabulary Object Detection without Human Curated Data Generation

MirrorShield: Towards Universal Defense Against Jailbreaks via Entropy-Guided Mirror Crafting

HICD: Hallucination-Inducing via Attention Dispersion for Contrastive Decoding to Mitigate Hallucinations in Large Language Models

RouterEval: A Comprehensive Benchmark for Routing LLMs to Explore Model-level Scaling Up in LLMs

Language Models, Graph Searching, and Supervision Adulteration: When More Supervision is Less and How to Make More More

Cost-Optimal Grouped-Query Attention for Long-Context Modeling

Beyond Matryoshka: Revisiting Sparse Coding for Adaptive Representation

On the Vulnerability of Concept Erasure in Diffusion Models

Char-mander Use mBackdoor! A Study of Cross-lingual Backdoor Attacks in Multilingual LLMs

SQLong: Enhanced NL2SQL for Longer Contexts with LLMs

DiffSampling: Enhancing Diversity and Accuracy in Neural Text Generation

TreeCut: A Synthetic Unanswerable Math Word Problem Dataset for LLM Hallucination Evaluation

Attention Mechanism for LLM-based Agents Dynamic Diffusion under Information Asymmetry

Robust Adaptation of Large Multimodal Models for Retrieval Augmented Hateful Meme Detection

R2-KG: General-Purpose Dual-Agent Framework for Reliable Reasoning on Knowledge Graphs

DeepResonance: Enhancing Multimodal Music Understanding via Music-centric Multi-way Instruction Tuning

MomentSeeker: A Task-Oriented Benchmark For Long-Video Moment Retrieval

EquiBench: Benchmarking Large Language Models' Understanding of Program Semantics via Equivalence Checking

EssayJudge: A Multi-Granular Benchmark for Assessing Automated Essay Scoring Capabilities of Multimodal Large Language Models

Uncovering Untapped Potential in Sample-Efficient World Model Agents

MMUnlearner: Reformulating Multimodal Machine Unlearning in the Era of Multimodal Large Language Models

CoLA: Compute-Efficient Pre-Training of LLMs via Low-Rank Activation

AI-driven Personalized Privacy Assistants: a Systematic Literature Review

Early Risk Prediction of Pediatric Cardiac Arrest from Electronic Health Records via Multimodal Fused Transformer

Online Scheduling for LLM Inference with KV Cache Constraints

Uni-Retrieval: A Multi-Style Retrieval Framework for STEM's Education

Speculative Prefill: Turbocharging TTFT with Lightweight and Training-Free Token Importance Estimation

Latent Action Learning Requires Supervision in the Presence of Distractors

Redefining Machine Unlearning: A Conformal Prediction-Motivated Approach

On the Role of Transformer Feed-Forward Layers in Nonlinear In-Context Learning

People who frequently use ChatGPT for writing tasks are accurate and robust detectors of AI-generated text

NBDI: A Simple and Effective Termination Condition for Skill Extraction from Task-Agnostic Demonstrations

MMDocIR: Benchmarking Multi-Modal Retrieval for Long Documents

Building Symbiotic AI: Reviewing the AI Act for a Human-Centred, Principle-Based Framework

TiEBe: Tracking Language Model Recall of Notable Worldwide Events Through Time

xLSTM-SENet: xLSTM for Single-Channel Speech Enhancement

A Separable Self-attention Inspired by the State Space Model for Computer Vision

Cross-model Transferability among Large Language Models on the Platonic Representations of Concepts

KunServe: Efficient Parameter-centric Memory Management for LLM Serving

Hotspot-Driven Peptide Design via Multi-Fragment Autoregressive Extension

Can LLMs be Good Graph Judge for Knowledge Graph Construction?

RoCoDA: Counterfactual Data Augmentation for Data-Efficient Robot Learning from Demonstrations

Rate, Explain and Cite (REC): Enhanced Explanation and Attribution in Automatic Evaluation by Large Language Models

Knowledge-Guided Prompt Learning for Request Quality Assurance in Public Code Review

Scaling Stick-Breaking Attention: An Efficient Implementation and In-depth Study

M-RewardBench: Evaluating Reward Models in Multilingual Settings

Unlearning Backdoor Attacks for LLMs with Weak-to-Strong Knowledge Distillation

RATE: Causal Explainability of Reward Models with Imperfect Counterfactuals

Bayes Adaptive Monte Carlo Tree Search for Offline Model-based Reinforcement Learning

Large Continual Instruction Assistant

TopoTune : A Framework for Generalized Combinatorial Complex Neural Networks

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment

Created by

Haebom

저자

Kun Wang, Guibin Zhang, Zhenhong Zhou, Jiahao Wu, Miao Yu, Shiqian Zhao, Chenlong Yin, Jinhu Fu, Yibo Yan, Hanjun Luo, Liang Lin, Zhihao Xu, Haolang Lu, Xinye Cao, Xinyun Zhou, Weifei Jin, Fanci Meng, Junyuan Mao, Yu Wang, Hao Wu, Minghe Wang, Fan Zhang, Junfeng Fang, Wenjie Qu, Yue Liu, Chengwei Liu, Yifan Zhang, Qiankun Li, Chongye Guo, Yalan Qin, Zhaoxin Fan, Yi Ding, Donghai Hong, Jiaming Ji, Yingxin Lai, Zitong Yu, Xinfeng Li, Yifan Jiang, Yanhui Li, Xinyu Deng, Junlin Wu, Dongxia Wang, Yihao Huang, Yufei Guo, Jen-tse Huang, Qiufeng Wang, Wenxuan Wang, Dongrui Liu, Yanwei Yue, Wenke Huang, Guancheng Wan, Heng Chang, Tianlin Li, Yi Yu, Chenghao Li, Jiawei Li, Lei Bai, Jie Zhang, Qing Guo, Jingyi Wang, Tianlong Chen, Joey Tianyi Zhou, Xiaojun Jia, Weisong Sun, Cong Wu, Jing Chen, Xuming Hu, Yiming Li, Xiao Wang, Ningyu Zhang, Luu Anh Tuan, Guowen Xu, Jiaheng Zhang, Tianwei Zhang, Xingjun Ma, Jindong Gu, Xiang Wang, Bo An, Jun Sun, Mohit Bansal, Shirui Pan, Lingjuan Lyu, Yuval Elovici, Bhavya Kailkhura, Yaodong Yang, Hongwei Li, Wenyuan Xu, Yizhou Sun, Wei Wang, Qing Li, Ke Tang, Yu-Gang Jiang, Felix Juefei-Xu, Hui Xiong, Xiaofeng Wang, Dacheng Tao, Philip S. Yu, Qingsong Wen, Yang Liu

개요

본 논문은 대규모 언어 모델(LLM)의 안전성에 대한 종합적인 검토를 제공합니다. 기존 연구들이 LLM 수명주기의 특정 단계(배포 단계 또는 미세 조정 단계 등)에만 초점을 맞춘 것과 달리, 본 논문은 데이터 준비, 사전 훈련, 사후 훈련, 배포 및 최종 상용화를 포함하는 전체 LLM 수명주기에 걸친 "풀스택" 안전성 개념을 처음으로 제시합니다. 800편 이상의 논문을 검토하여 LLM 안전성 문제에 대한 포괄적인 이해를 제공하며, 데이터 생성, 정렬 기법, 모델 편집 및 LLM 기반 에이전트 시스템과 같은 유망한 연구 방향을 제시합니다.

시사점, 한계점

•

시사점:

◦

LLM의 전체 수명주기(데이터 준비부터 상용화까지)를 고려하는 "풀스택" 안전성 개념을 최초로 제시.

◦

800편 이상의 논문을 분석하여 LLM 안전성 문제에 대한 포괄적인 이해 제공.

◦

데이터 생성, 정렬 기법, 모델 편집, LLM 기반 에이전트 시스템 등의 유망한 연구 방향 제시.

◦

LLM 안전성 연구에 대한 종합적인 로드맵 및 관점 제공.

•

한계점:

◦

본 논문 자체가 제시하는 한계점에 대한 언급은 없음. 향후 연구를 위한 방향 제시에 집중.

Made with Slashpage