James Stokes

2 accepted papers

2019

Fisher-Rao Metric, Geometry, and Complexity of Neural Networks

Tengyuan Liang, Tomaso Poggio, Alexander Rakhlin, James Stokes

AISTATS 2019poster

We study the relationship between geometry and capacity measures for deep neural networks from an invariance viewpoint. We introduce a new notion of capacity — the Fisher-Rao norm — that possesses desirable invariance properties and is motivated by Information Geometry. We discover an analytical cha…

Cited by 274 Source ↗PDF ↗Scholar ↗

2019

Interaction Matters: A Note on Non-asymptotic Local Convergence of Generative Adversarial Networks

Tengyuan Liang, James Stokes

AISTATS 2019poster

Motivated by the pursuit of a systematic computational and algorithmic understanding of Generative Adversarial Networks (GANs), we present a simple yet unified non-asymptotic local convergence theory for smooth two-player games, which subsumes several discrete-time gradient-based saddle point dynami…

Cited by 249 Source ↗PDF ↗Scholar ↗