Machine Learning Lunch Meeting

General: MLLM is a cross-discipline weekly seminar open to all where UW-Madison professors present their research in machine learning, both theory and applications.  The goal is to promote the great work at UW-Madison and stimulate collaborations. Please see our website for more information.

Speaker: Ilias Diakonikolas (CS)

Abstract: Multi-index models (MIMs) are functions that depend on the projection onto a low-dimensional subspace. These models provide a useful framework for analyzing a wide range of machine learning problems, such as multiclass linear classification, learning intersections of halfspaces, and more complex neural networks. Despite extensive investigation, there remains a significant gap in our understanding of the efficient learnability of MIMs.

In this talk, we will survey recent algorithmic work on learning MIMs, focusing on methods with provable performance guarantees. In particular, we will present a robust noise-tolerant learning algorithm that works for a broad class of well-behaved MIMs, under standard distributional assumptions. As applications, we will demonstrate how this framework leads to much faster noise-tolerant learning algorithms for multiclass linear classifiers and intersections of halfspaces.