报告题目(Title)：Dependable Intelligence and Reasoning Beyond LLM

时间 (Date & Time)：2024.10.08 16:00-18:00

地点 (Location)：理科一号楼1126报告厅（燕园校区） Lecture hall 1126, Science Building #1 (Yanyuan)

主讲人 (Speaker)：Jin Song Dong (National University of Singapore/新加坡国立大学)

邀请人 (Host)：Sergey Mechtaev

报告摘要 (Abstract)：

Machine Learning (ML) systems have become increasingly integral to safety and security-critical applications. However, a significant challenge arises from the inherent lack of explainability and verifiability in many ML systems. Our recent research has focused on addressing this issue by developing a Trusted ML system. The initial segment of this presentation delves into the "Silas: Trusted Machine Learning System," an initiative that seamlessly integrates open machine learning with formal automated reasoning (www.depintel.com). In the subsequent part of the discussion, we explore the reasoning capabilities of LLM (encompassing ChatGPT3.5 and GPT4). Specifically, we discuss the approaches to link LLM with formal reasoning techniques, aiming to establish a framework for trusted LLM agents. As a practical demonstration, we will present the application of probabilistic reasoning, machine learning, LLM, and computer vision to sports analytics and share the vision of a new international sports analytics conference series (https://formal-analysis.com/isace/2025/).

主讲人简介(Bio)：

Jin Song Dong is a professor at the National University of Singapore. His research interests include safety and security systems, sports analytics, and trusted machine learning/LLM reasoning. He co-founded the commercialized PAT verification system which has garnered thousands of registered users from over 150 countries. Jin Song co-founded the commercialized trusted machine learning system Silas. He has received numerous best paper awards and served on the editorial board of ACM Transactions on Software Engineering and Methodology and Formal Aspects of Computing. He has successfully supervised 30 PhD students and is an Australian Institute of Engineering Fellow. In his leisure time, Jin Song developed Markov Decision Process models for tennis analysis using PAT, assisting professional players with pre-match analysis (beating the world's best). He is a Junior Grand Slam coach and coached tennis to his three children, all of whom have reached the #1 national junior ranking in Singapore/Australia. Two of his children have earned US NCAA full scholarships. His second son, Chen, played #1 singles for Australia in the Junior Davis Cup Final and participated in both the Australian Open and US Open Junior Grand Slams.

欢迎关注计算机学院微信公众号，了解更多讲座信息！

北京大学计算机学院