Abstract: This talk delves into the evolution and understanding of AI systems through the lens of mathematical functions. The genesis of modern AI systems is rooted deeply in years of trial and error, which have led to cutting-edge deep learning architectures. In recent years, novel mathematical frameworks have emerged to dissect these systems further, enhancing our understanding of their fundamental workings. With a focus on our recent work on ‘Radon-Domain Bounded Variation Spaces’, the talk illuminates how these innovative function spaces unravel key elements contributing to the current success of AI systems. Further, the talk explores how these discoveries pave the path for a rigorous understanding and improvement of AI. Ultimately, the aim is to harness these insights and theories to engineer AI systems mirroring the robustness, reliability, and interpretability of contemporary communication systems. This talk is based on joint work with Rahul Parhi, Joe Shenouda, and Liu Yang.
Bio: Robert Nowak holds the Keith and Jane Nosbusch Professorship in Electrical and Computer Engineering at the University of Wisconsin-Madison, where he directs the AFOSR/AFRL University Center of Excellence on Data Efficient Machine Learning. His research focuses on machine learning, optimization, and signal processing. He serves on the editorial boards of the SIAM Journal on the Mathematics of Data Science and the IEEE Journal on Selected Areas in Information Theory.