Learning on graphs is a long-standing and fundamental challenge in machine learning and recent works have demonstrated solid progress in this area. However, most existing models tacitly assume a closed-world setting where the training and testing data are sampled from an identical distribution. In this talk, we will introduce recent advances that develop theoretically principled and practical useful methods for learning on graphs in the challenging open-world hypothesis, where the model needs to generalize to out-of-distribution testing data in the wild. From the methodological view, we will present two technical paths for building provably generalizable learning algorithms, based on invariance and causality principles, respectively. On the applicable side, we will discuss how to apply these methods to address the pressing problems in recommender systems and molecular discovery.
Speaker Bio
Qitian Wu is a postdoctoral fellow at Broad Institute of MIT and Harvard. Prior to this, he achieved PhD in Computer Science at Shanghai Jiao Tong University. His research interest focuses on machine learning with complex structured data. His recent works endeavor to develop efficient foundational backbones for representing large-scale graph data and provably generalizable learning algorithms for handling distribution shifts. He also seek to apply this methodology to address the pressing problems in recommender systems and biomedical science. He is the recipient of Eric and Wendy Schmidt Center Fellowship, Microsoft Research Fellowship, Baidu Scholarship and Rising Star in Artificial Intelligence.
More Details
- When: Wed 27 Nov 2024, at 1 - 2 pm (Brisbane time)
- Speaker: Dr Qitian Wu (MIT & Harvard)
- Host: Dr Ruihong Qiu
- Venue: Online
- Zoom: https://uqz.zoom.us/j/85331921600 [Recording]
No.24-15 Evaluation and Reasoning in Real-world Scenarios