Events

07. September 2023, 15:00 until 17:00
Seminar: Alternating minimization and some generalizations of convexity, Pierre-Cyril Aubin, TU Wien
Abstract:
I will present my recent research in going beyond the quadratic cost in optimization, by replacing it with a general cost function. With Flavien Léger (INRIA Paris), we unveiled in https://arxiv.org/abs/2305.04917 a new class of gradient-type optimization methods that extends vanilla gradient descent, mirror descent, Riemannian gradient descent, and natural gradient descent. Our approach involves constructing a surrogate for the objective function in a systematic manner, based on a chosen cost function. This surrogate is then minimized using an alternating minimization scheme. Using optimal transport theory we establish convergence rates based on generalized notions of smoothness and convexity. We provide local versions of these two notions when the cost satisfies a condition known as nonnegative cross-curvature. In particular our framework provides the first global rates for natural gradient descent and Newton's method. For a quick read of https://arxiv.org/abs/2305.04917 I recommend reading the summary p4-6 and then look for your favorite algorithm (mirror, Riemannian, etc) in Section 4.