Date: Thursday, September 10, 2026
Time: 2:00 PM
Location: DC 1304
Title: Can NNs Learn Computational Algorithms That Generalize OOD? Perspectives from the Interplay Between Architecture and Computational Structure
Abstract: A central challenge in modern machine learning is learning generalizable procedures that remain effective on unseen, potentially out-of-distribution (OOD) data. Such generalization depends on a complex interplay among model architectures, task structures, data assumptions, and training methodologies. In this talk, Yusu will focus on the interaction between model architecture and task structure in the context of graphs tasks or geometric problems. We are particularly interested in the following questions: Do different neural networks learn fundamentally different algorithmic procedures? Can OOD generalization be achieved with only finite samples? How do we probe what's learned internally? How can we use obtained insights help design more effective neural models that can tackle computationally hard (geometric) problems more efficiently? Yusu will present some of these initial studies exploring these questions. This talk is based on joint work with several collaborators, whom Yusu will acknowledge during the talk.
Bio: Yusu Wang is an HDSI Endowed Chair Professor in the Halıcıoğlu Data Science Institute at the University of California San Diego, where she holds an affiliated faculty position in the Department of Computer Science and Engineering. She serves as the Director of the NSF National AI Institute TILOS (The Institute for Learning Enabled Optimization at Scale), a collaborative effort leading researchers across UCSD, MIT, National University, UPenn, UT Austin, and Yale. Professor Wang received her B.S degree with First Class Honours from Tsinghua University and earned her M.S and Ph.D degrees from Duke University, winning the Best PhD Dissertation Award in Computer Science. She was a postdoctoral researcher in the Geometric Computing Lab at Stanford University before joining the faculty at The Ohio State University as a Professor of Computer Science and Engineering, where she co-directed the Foundations of Data Science Research Community of Practice at the Translational Data Analysis Institute.