Department of Electrical and Systems Engineering
University of Pennsylvania
A picture of the energy landscape of deep neural networks
Deep networks are mysterious. These over-parametrized machine learning models, trained with rudimentary optimization algorithms on non-convex landscapes in millions of dimensions have defied attempts to put a sound theoretical footing beneath their impressive performance.
This talk will shed light upon some of these mysteries. I will employ diverse ideas — from thermodynamics and optimal transportation to partial differential equations, control theory and Bayesian inference — and paint a picture of the training process of deep networks. Along the way, I will develop state-of-the-art algorithms for non-convex optimization. The goal of machine perception is not just to classify objects in images but instead, enable intelligent agents that can seamlessly interact with our physical world. I will conclude with a vision of how advances in machine learning and robotics may come together to help build such an Embodied Intelligence.
A pizza lunch will be served.