PhD Dissertation Proposal by HE Junda | From Single-Agent Reliability to Multi-Agent Synergy: A Software Engineering Perspective towards Deployable Autonomous Agents

From Single-Agent Reliability to Multi-Agent Synergy: A Software Engineering Perspective towards Deployable Autonomous Agents

ABOUT THE TALK

Autonomous agents, powered by Deep Reinforcement Learning (DRL) and Large Language Models (LLMs), hold immense promise. However, their transition from controlled simulations to the real world is hindered by critical challenges in reliability. Ensuring these agents are safe, effective, and engaging in productive teamwork, requires a rigorous software engineering approach that goes beyond traditional evaluation methods. This dissertation addresses this gap by developing a suite of novel frameworks for the comprehensive testing of autonomous agents. We address agent safety, policy optimality. Building on this foundation of single-agent reliability, we extend our focus to multi-agent collaboration. We provide a systematic review of LLM-Based Multi-Agent (LMA) systems in software engineering and propose a comprehensive research agenda to guide their future development.

SPEAKER BIOGRAPHY

Junda HE is a Ph.D. candidate in Computer Science at the School of Computing and Information Systems, Singapore Management University, under the supervision of Professor David LO. His research focuses on various aspects of Large Language Models for Software Engineering (LLM4SE) and Trustworthy AI, with outcomes published in premier venues such as ICSE, TOSEM, TSE, and ASE. He also serves as a reviewer for leading journals and conferences, including Communications of the ACM (CACM), IEEE Transactions on Software Engineering (TSE), ACM Transactions on Software Engineering and Methodology (TOSEM), ACL, and UIST. In recognition of his academic achievements, he has been awarded the SMU Presidential Doctoral Fellowship. Outside of his academic pursuits, he enjoys participating in outdoor sports.