Program / GS-04: LLM Applications, Reasoning & Multi-Agent Systems

GS-04: LLM Applications, Reasoning & Multi-Agent Systems

November 05 (Thursday) · 10:20–12:20 · Room 2

Chair: Takeru Aoki (The University of Tokyo)

Back to timetable

10:20–10:40Paper ID: 81Full paper

Detecting Adversarial Agents in Multi-Agent Debate via Token-Confidence Margin

Insung Hwang (Sungkyunkwan University)*; Cheolwon Na (Sungkyunkwan University); Jee-Hyong Lee (Sungkyunkwan University)

10:40–11:00Paper ID: 118Full paper

Question Answering using Hierarchy-Aware Retrieval-Augmented Generation for Sports Competition Rules

Shion Kawaguchi (Hiroshima City University)*; Keiichi Tamura (Hiroshima City University)

11:00–11:20Paper ID: 149Full paper

Comparing Problem-Specific and Mean Hidden-State Interventions for Context-Reconstructed Reasoning

Sena Sasahara (Kyushu Institute of Technology)*; Kazutaka Shimada (Kyushu Institute of Technology)

11:20–11:40Paper ID: 184Full paper

A Deep Learning Generative Model for Automatic Role-Play Prompt Generation

Mahardika Bayu Setya Budi (Telkom University); Andi Yunus (Telkom University)*

11:40–11:55Paper ID: 197Late Breaking

A Genetic-Fuzzy Consensus Framework for Multi-Agent LLM Differential Diagnosis Under Uncertainty

Nafis Naufal Rahman (Universitas Brawijaya)*; Dionisius Seraf Saputra (Universitas Brawijaya)

Presentations finish at 11:55. The scheduled session ends at 12:20.

Full paper: 15 minutes presentation + 5 minutes Q&A. Short, abstract-only and late breaking papers: 10 minutes presentation + 5 minutes Q&A. Speakers should bring their own laptop; presentation PCs are not provided.