Towards Reliable Task-oriented LLM Agents
About this Topic:
Large language model agents have gained immense attention for general task intelligence. However, their probabilistic nature causes unintended behaviors and execution uncertainty, leading to poor real-world performance. Traditional approaches focus on mitigating hallucinations to reduce uncertainty. In contrast, this lecture, from cognitive conversational interaction, takes "irreducible uncertainty" as the premise. It redefines task-agent execution reliability, explores uncertainty modeling, and introduces a novel uncertainty-aware reliability alignment paradigm for task-oriented LLM agents.
About the Presenter:
Kai Yu (F'26) received the B.Eng. and M.Sc. degrees from Tsinghua University, China, and the Ph.D. degree from the University of Cambridge, U.K. in 1999, 2002 and 2006, respectively.
He is currently a Distinguished Professor and the Director of the Machine Intelligence Institute at Shanghai Jiao Tong University (SJTU), and Co-founder and Chief Scientist of AISpeech. He previously worked at the University of Cambridge, U.K as a senior research associate. He returned to China and joined Shanghai Jiao Tong University (SJTU) in 2012. His research interests lie in the field of conversational AI, including rich aspects of speech and language processing with the goal of building cognitive conversational agent capable of operating in complex real-world environment and evolving via interacting with environment.
Dr. Yu has published over 300 peer-reviewed papers and won numerous paper awards. He founded AISpeech to commercialize state-of-the-art conversational AI technology, which has been selected into the “AI Key Players” list in the Equity Research Report of AI by Goldman Sachs and one of the Cool Vendors for AI (East Asia) by Gartner. He is a fellow of IEEE and a fellow of International Speech Communication Association (ISCA). He previously served on the IEEE Speech and Language Processing Technical Committee (2017–2019) and as Associate Editor of IEEE/ACM Transactions on Audio, Speech, and Language Processing (2019–2024). He is currently a Board Member of the IEEE Signal Processing Society Conferences Board and Membership Board. He also serves as the Director of the Speech, Dialogue and Auditory Processing Technical Committee of the China Computer Federation (CCF) and is a member of the CCF Council.
Want to learn more about upcoming events & webinars?
Visit the events section of the Signal Processing website to see all upcoming lectures, workshops, webinars, and more.