TechPulseTelegram
โ† PrevSat, Feb 21Next โ†’

๐Ÿค– AI Agent23 items

Latest AI agent research papers, analysis, and improvement insights

arxivcs.LG

MASPO: Unifying Gradient Utilization, Probability Mass, and Signal Reliability for Robust and Sample-Efficient LLM Reasoning

LLM ์ถ”๋ก ์„ ์œ„ํ•œ ๊ฒฌ๊ณ ํ•œ ๊ทธ๋ž˜๋””์–ธํŠธ ํ™œ์šฉ ์ตœ์ ํ™”

arxivcs.AI

AutoNumerics: An Autonomous, PDE-Agnostic Multi-Agent Pipeline for Scientific Computing

๊ณผํ•™ ๊ณ„์‚ฐ์šฉ PDE ํ•ด๊ฒฐ ๋‹ค์ค‘ ์—์ด์ „ํŠธ ์ž๋™ ์„ค๊ณ„

arxivcs.CL

Modeling Distinct Human Interaction in Web Agents

์›น ์—์ด์ „ํŠธ์˜ ์ธ๊ฐ„ ๊ฐœ์ž… ํŒจํ„ด ๋ชจ๋ธ๋ง ๋ฐ ํ˜‘๋ ฅ

arxivcs.AI

KLong: Training LLM Agent for Extremely Long-horizon Tasks

๊ถค์  ๋ถ„ํ•  SFT์™€ ์ ์ง„์  RL๋กœ ์žฅ๊ธฐ ์ž‘์—… LLM ์—์ด์ „ํŠธ ํ›ˆ๋ จ

semantic_scholarresearch

MedClarify: An information-seeking AI agent for medical diagnosis with case-specific follow-up questions

๋ฐ˜๋ณต์  ์งˆ๋ฌธ์œผ๋กœ ์˜๋ฃŒ ์ง„๋‹จ ๋ถˆํ™•์‹ค์„ฑ ํ•ด๊ฒฐํ•˜๋Š” ์—์ด์ „ํŠธ

semantic_scholarresearch

KLong: Training LLM Agent for Extremely Long-horizon Tasks

๊ถค์  ๋ถ„ํ•  SFT์™€ ์ ์ง„์  RL๋กœ ์žฅ๊ธฐ ์ž‘์—… LLM ์—์ด์ „ํŠธ ํ›ˆ๋ จ

semantic_scholarresearch

What Makes a Good LLM Agent for Real-world Penetration Testing?

์นจํˆฌ ํ…Œ์ŠคํŠธ LLM ์—์ด์ „ํŠธ์˜ ์‹คํŒจ ์›์ธ ๋ถ„์„

arxivcs.LG

FAMOSE: A ReAct Approach to Automated Feature Discovery

ReAct ๊ธฐ๋ฐ˜ ์ž๋™ ํŠน์„ฑ ๊ณตํ•™ ํ”„๋ ˆ์ž„์›Œํฌ

arxivcs.LG

When to Trust the Cheap Check: Weak and Strong Verification for Reasoning

LLM ์ถ”๋ก ์˜ ์•ฝ์•ฝ/๊ฐ• ๊ฒ€์ฆ ์‹ ๋ขฐ๋„ ๋ฐ ๋น„์šฉ ๋ถ„์„

arxivcs.LG

Stable Asynchrony: Variance-Controlled Off-Policy RL for LLMs

LLM ๋น„๋™๊ธฐ ๊ฐ•ํ™”ํ•™์Šต์˜ ๋ถ„์‚ฐ ์ œ์–ด ๊ธฐ๋ฒ•

arxivcs.AI

Evaluating Chain-of-Thought Reasoning through Reusability and Verifiability

๋‹ค์ค‘ ์—์ด์ „ํŠธ CoT ์ถ”๋ก ์˜ ์žฌ์‚ฌ์šฉ์„ฑ๊ณผ ๊ฒ€์ฆ ํ‰๊ฐ€

semantic_scholarresearch

NeuDiff Agent: A Governed AI Workflow for Single-Crystal Neutron Crystallography

๊ฒฐ์ •ํ•™ ์ž๋™ ๋ถ„์„ ๋„๊ตฌ ์‚ฌ์šฉ AI ์›Œํฌํ”Œ๋กœ์šฐ

semantic_scholarresearch

Testing BDI-based Multi-Agent Systems using Discrete Event Simulation

BDI ๋‹ค์ค‘ ์—์ด์ „ํŠธ ์‹œ์Šคํ…œ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ํ…Œ์ŠคํŠธ

arxivcs.AI

AI Gamestore: Scalable, Open-Ended Evaluation of Machine General Intelligence with Human Games

์ธ๊ฐ„ ๊ฒŒ์ž„์œผ๋กœ AI ์ผ๋ฐ˜ ์ง€๋Šฅ ํ‰๊ฐ€ ๋ฒค์น˜๋งˆํฌ

arxivmath.OC

Adaptive Decentralized Composite Optimization via Three-Operator Splitting

๋„คํŠธ์›Œํฌ ๋ถ„์‚ฐ ์—์ด์ „ํŠธ์˜ ์ ์‘์  ์ตœ์ ํ™” ์•Œ๊ณ ๋ฆฌ์ฆ˜

arxivphysics.acc-ph

Toward a Fully Autonomous, AI-Native Particle Accelerator

์ž์œจ ์ž…์ž๊ฐ€์†๊ธฐ ์šด์˜์„ ์œ„ํ•œ AI ๊ณต๋™ ์„ค๊ณ„

semantic_scholarresearch

Unveiling the augmented human agent: How communication style mitigates the heuristic barriers of human-AI collaboration in online customer service encounters

๊ณ ๊ฐ ์„œ๋น„์Šค์˜ ์ธ๊ฐ„-AI ํ˜‘๋ ฅ ์ปค๋ฎค๋‹ˆ์ผ€์ด์…˜ ์—ฐ๊ตฌ

blog_openairesearch

Our First Proof submissions

์ˆ˜ํ•™ ์ฆ๋ช… ์ฑŒ๋ฆฐ์ง€์—์„œ AI ๋ชจ๋ธ์˜ ์ถ”๋ก  ๋Šฅ๋ ฅ ์‹œ์—ฐ

arxivcs.AI

CLEF HIPE-2026: Evaluating Accurate and Efficient Person-Place Relation Extraction from Multilingual Historical Texts

๋‹ค๊ตญ์–ด ์—ญ์‚ฌ ํ…์ŠคํŠธ์—์„œ ์ธ๋ฌผ-์žฅ์†Œ ๊ด€๊ณ„ ์ถ”์ถœ ํ‰๊ฐ€

arxivcs.CL

Unmasking the Factual-Conceptual Gap in Persian Language Models

ํŽ˜๋ฅด์‹œ์•„ ์–ธ์–ด ๋ชจ๋ธ์˜ ์‚ฌ์‹ค-๊ฐœ๋… ๊ฒฉ์ฐจ ์ง„๋‹จ

Page 1 of 2Next