Learning How to Remember: Memory Modality Selection for GUI Agents
Text or image memory, selected for each GUI interaction step.
Research
Papers on language agents, multimodal reasoning, and learning systems by Taeyun Roh (노태윤).
Text or image memory, selected for each GUI interaction step.
Training-free contrastive scoring reduces text-induced bias in multimodal multiple-choice QA.
Step-aware reinforcement learning reduces cascading errors in medical multimodal reasoning.
Expert planning distillation and dynamic search for reimbursement-aware diabetes prescribing.
Agent-driven clustering organizes memory for small language model agents.
분할-연합학습의 데이터, 자원 이질성 극복을 위한 연구 동향 분석
A survey of data and resource heterogeneity in split-federated learning.
허가형 블록체인을 위한 보안 컨테이너 활용 방안 연구
Secure containers combine isolation and lightweight deployment for permissioned blockchain nodes.