Skip to content

Asian Agent Research (日韓)

WorkYearKey Contribution
AI Scientist-v22025First fully AI-generated peer-reviewed paper accepted at ICLR workshop; agentic tree search for autonomous scientific discovery
CycleQD2025 (ICLR)Evolutionary algorithm for acquiring multiple agentic skills in LLMs via quality diversity + model merging
Evolutionary Model Merge2024Nature-inspired evolution for combining open-source models; published in Nature Machine Intelligence
ShinkaEvolve2025LLM-driven evolutionary framework; SOTA on circle packing, AIME reasoning
M2N22025 (GECCO)Nature-inspired competition/attraction for model merging niches
WorkFocus
Deep Learning + Autonomous AgentsCore AI research, applied to robotics and industrial automation
WorkFocus
LLM agent architecturesActive research on agent reasoning, tool use, multi-agent coordination
Multi-agent systems + AI reasoningKyoto University long-standing MAS research tradition
WorkYearContribution
Agentic RL Survey (日本語)2025Comprehensive Japanese survey covering 500+ papers on RL for agentic LLM capabilities

WorkYearKey Contribution
FluidGPT2025 (MobiCom)UI-based autonomous mobile agent for app control; demonstrated at ACM MobiCom
AI Semiconductor ecosystemOngoingKAIST + Naver + Intel collaboration on Gaudi optimization
Multi-agent learningOngoingDistributed multi-agent systems research
WorkFocus
HyperCLOVALarge-scale Korean foundation model
Agent NAI agent platform built on HyperCLOVA
Enterprise AI agentsAutonomous workflow agents
WorkFocus
LLM agentsAutonomous agent development
Multi-agent workflowsEnterprise agent orchestration
WorkFocus
Foundation modelsLarge-scale Korean LLM development
Agent systemsAutonomous agent architecture research

WorkYearContribution
LangGraph + CrewAI Hybrid (KSCI)2025Integration of role-based CrewAI with graph-based LangGraph for Korean-language MAS
AgentsNet (한국어)2025Multi-agent coordination benchmark up to 100 agents; self-organization metrics
NAMU Agent Platform2025Enterprise-grade autonomous agent workflow with 5-stage agentic workflow
AIOS: LLM Agent OS (한국어 blog)2024Korean-language summary of LLM-as-OS-kernel concept

Papers at NeurIPS, ICML, ICLR, EMNLP with Japanese/Korean first authors:

WorkVenueYearContribution
RAPNeurIPS 20242024Multimodal agents with past experience retrieval for planning
AVATARNeurIPS 20242024Comparator module for automated prompt optimization in tool use
HIAGENTNeurIPS 20242024Subgoal-based memory chunking for long-horizon tasks
Optimus-1NeurIPS 20242024KG + experience pool for Minecraft agents
Tulip AgentICML 20242024Extensible tool library with vector search
Agent QICLR 20252025MCTS + DPO for web agent training; 340% improvement
ADASICLR 20252025Meta agent search for automatic agent architecture discovery
AGENTFLOWICLR 20252025Trainable in-the-flow agentic framework with Flow-GRPO
Agentic Self-LearningICLR 20252025Multi-role closed-loop RL without human supervision

CountryInstitutionSpecialization
🇯🇵Sakana AIEvolutionary AI, agentic systems, model merging
🇯🇵Preferred NetworksDeep learning, autonomous agents
🇯🇵University of TokyoLLM agent architectures
🇯🇵Kyoto UniversityMulti-agent systems, AI reasoning
🇰🇷KAISTMobile agents, AI semiconductors, multi-agent learning
🇰🇷NaverHyperCLOVA, Agent N, enterprise agents
🇰🇷Kakao AILLM agents, autonomous workflows
🇰🇷LG AI ResearchFoundation models, agent systems