TechPulseTelegram
← PrevMon, Apr 13Next β†’

πŸ€– AI Agent15 items

Latest AI agent research papers, analysis, and improvement insights

semantic_scholarresearch

Human Values Matter: Investigating How Misalignment Shapes Collective Behaviors in LLM Agent Communities

닀쀑 μ—μ΄μ „νŠΈ μ‹œμŠ€ν…œμ˜ 인간 κ°€μΉ˜ μ •λ ¬ λΆ€μž¬κ°€ 집단 행동에 λ―ΈμΉ˜λŠ” 영ν–₯ 연ꡬ

semantic_scholarresearch

Lightweight LLM Agent Memory with Small Language Models

μ†Œν˜• μ–Έμ–΄λͺ¨λΈ 기반 κ²½λŸ‰ λ©”λͺ¨λ¦¬λ‘œ μž₯κΈ° μƒν˜Έμž‘μš© 일관성 μœ μ§€

semantic_scholarresearch

SuperLocalMemory V3.3: The Living Brain -- Biologically-Inspired Forgetting, Cognitive Quantization, and Multi-Channel Retrieval for Zero-LLM Agent Memory Systems

생물학 영감 인지 ν”„λ‘œμ„ΈμŠ€ 기반 둜컬 μ—μ΄μ „νŠΈ λ©”λͺ¨λ¦¬ μ‹œμŠ€ν…œ κ΅¬ν˜„

semantic_scholarresearch

FLARE: Agentic Coverage-Guided Fuzzing for LLM-Based Multi-Agent Systems

닀쀑 μ—μ΄μ „νŠΈ LLM μ‹œμŠ€ν…œμ˜ λ¬΄ν•œ 루프, μ‹€νŒ¨ 감지 퍼징 ν”„λ ˆμž„μ›Œν¬

semantic_scholarresearch

Dynamic Attentional Context Scoping: Agent-Triggered Focus Sessions for Isolated Per-Agent Steering in Multi-Agent LLM Orchestration

닀쀑 μ—μ΄μ „νŠΈ μ˜€μΌ€μŠ€νŠΈλ ˆμ΄μ…˜μ˜ μ»¨ν…μŠ€νŠΈ μ˜€μ—Ό ν•΄κ²° DACS λ©”μ»€λ‹ˆμ¦˜

semantic_scholarresearch

Your LLM Agent Can Leak Your Data: Data Exfiltration via Backdoored Tool Use

νŒŒμΈνŠœλ‹ LLM μ—μ΄μ „νŠΈμ˜ 도ꡬ μ‚¬μš©μ„ μ•…μš©ν•œ 데이터 유좜 곡격 뢄석

semantic_scholarresearch

Program Analysis Guided LLM Agent for Proof-of-Concept Generation

μ½”λ“œ 뢄석 기반 LLM μ—μ΄μ „νŠΈμ˜ μžλ™ 취약점 PoC 생성

semantic_scholarresearch

How Much LLM Does a Self-Revising Agent Actually Need?

μžκΈ°μˆ˜μ • μ—μ΄μ „νŠΈμ˜ μ—­λŸ‰ 쀑 LLMκ³Ό λͺ…μ‹œμ  ꡬ쑰의 기여도 뢄석

semantic_scholarresearch

Multi-Agent LLM Framework for Autonomous Network Fault Remediation

λ„€νŠΈμ›Œν¬ 결함 μžλ™ 탐지·진단·해결 μœ„ν•œ 닀쀑 μ—μ΄μ „νŠΈ RAG ν”„λ ˆμž„μ›Œν¬

semantic_scholarresearch

TOOLCAD: Exploring Tool-Using Large Language Models in Text-to-CAD Generation with Reinforcement Learning

도ꡬ μ‚¬μš© LLM의 CAD 엔진 μƒν˜Έμž‘μš© μ΅œμ ν™” ν…μŠ€νŠΈ-CAD μ—μ΄μ „νŠΈ

semantic_scholarresearch

TurboAgent: An LLM-Driven Autonomous Multi-Agent Framework for Turbomachinery Aerodynamic Design

터보기계 곡기역학 섀계 μžλ™ν™”λ₯Ό μœ„ν•œ 닀쀑 μ—μ΄μ „νŠΈ LLM ν”„λ ˆμž„μ›Œν¬

semantic_scholarresearch

ReRec: Reasoning-Augmented LLM-based Recommendation Assistant via Reinforcement Fine-tuning

κ°•ν™”ν•™μŠ΅ νŒŒμΈνŠœλ‹μœΌλ‘œ μΆ”μ²œ μž‘μ—… μΆ”λ‘  λŠ₯λ ₯ κ°•ν™”ν•œ LLM μ—μ΄μ „νŠΈ

hackernewsresearch

MiniMax M2.7 Is Now Open Source

MiniMax M2.7 μ—μ΄μ „νŠΈ λͺ¨λΈ μ˜€ν”ˆμ†ŒμŠ€ 곡개

semantic_scholarresearch

Reinforcement Learning with LLM-Guided Action Spaces for Synthesizable Lead Optimization

ν•©μ„± κ°€λŠ₯μ„± μ œμ•½ 기반 μ‹ μ•½ λΆ„μž μ΅œμ ν™” κ°•ν™”ν•™μŠ΅ μ—μ΄μ „νŠΈ

hackernewsresearch

Wikipedia's AI agent row likely just the beginning of the bot-ocalypse

μœ„ν‚€ν”Όλ””μ•„ AI μ—μ΄μ „νŠΈ λ…ΌμŸκ³Ό μžλ™ν™” λ΄‡μ˜ 미래 μ •μ±… 이슈