The Data Science Discipline of the School of EECS is hosting the following guest seminar:
Towards Personalized AI Agent Systems
Speaker: Jiaqi Zhang
Host: Dr. Ruihong Qiu
Host: Dr. Ruihong Qiu
Abstract:
Intelligent agents are increasingly expected to interact with complex digital and physical environments, yet most existing systems remain largely task-oriented and overlook individual user needs. This talk explores personalized agentic reasoning, where agents must understand users from ongoing interactions and incorporate personalized factors into their decisions and actions. We first introduce a personalized agent environment designed to capture explicit user factors. We then present Chain-of-User-Thought (COUT), which organizes personalized reasoning into progressively connected stages from environment understanding to user-need inference and personalized decision-making. Finally, we discuss how episode-level learning can help agents capture long-horizon dependencies across interactions. Together, these efforts aim to move agentic systems beyond task completion toward more adaptive and personalized human–agent interaction.
Intelligent agents are increasingly expected to interact with complex digital and physical environments, yet most existing systems remain largely task-oriented and overlook individual user needs. This talk explores personalized agentic reasoning, where agents must understand users from ongoing interactions and incorporate personalized factors into their decisions and actions. We first introduce a personalized agent environment designed to capture explicit user factors. We then present Chain-of-User-Thought (COUT), which organizes personalized reasoning into progressively connected stages from environment understanding to user-need inference and personalized decision-making. Finally, we discuss how episode-level learning can help agents capture long-horizon dependencies across interactions. Together, these efforts aim to move agentic systems beyond task completion toward more adaptive and personalized human–agent interaction.
Bio:
Jiaqi Zhang is a Ph.D. student at the School of Electrical Engineering and Computer Science, The University of Queensland, under the supervision of Prof. Hongzhi Yin and Assoc. Prof. Rocky Tong Chen. His research interests lie in Large Language Models, LLM Agents, and Recommender Systems.
Jiaqi Zhang is a Ph.D. student at the School of Electrical Engineering and Computer Science, The University of Queensland, under the supervision of Prof. Hongzhi Yin and Assoc. Prof. Rocky Tong Chen. His research interests lie in Large Language Models, LLM Agents, and Recommender Systems.
About Data Science Seminar
This seminar series is hosted by EECS Data Science.
Venue
Room: 78 - 631/632
Zoom link: https://uqz.zoom.us/j/86359949412
Zoom link: https://uqz.zoom.us/j/86359949412