Joan Bruna is an Associate Professor of Computer Science, Data Science and Mathematics (affiliated) at the Courant Institute and the Center for Data Science, New York University. He belongs to the CILVR (Computational Intelligence, Learning, Vision and Robotics) group and co-founded the MaD (Math and Data) group. His research interests touch several areas of Machine Learning, Signal Processing and High-Dimensional Statistics.
Talk: On Learning Sparse High-Dimensional Functions
Abstract: Neural Networks are hailed for their ability to discover useful low-dimensional features out of complex high-dimensional data. Over the recent years, the class of sparse (or 'multi-index') functions has emerged as a model with both practical motivations and rich structure, enabling a quantitative mathematical theory of 'feature learning'. In this talk, He will review recent progress on this front, by describing (i) the ability of gradient-descent algorithms to efficiently learn the multi-index class over Gaussian data, (ii) the robustness of such GD algorithms to non-Gaussian data, and (iii) SQ lower bounds for the single-index class. Joint work with Loucas Pillaud-Vivien, Aaron Zweig and Alex Damian.