Open Peer Review. Open Publishing. Open Access. Open Discussion. Open Directory. Open Recommendations. Open API. Open Source.
Stabilizing Adversarial Nets with Prediction Methods
Abhay Yadav, Sohil Shah, Zheng Xu, David Jacobs, Tom Goldstein
Feb 15, 2018 (modified: Feb 24, 2018)ICLR 2018 Conference Blind Submissionreaders: everyoneShow Bibtex
Abstract:Adversarial neural networks solve many important problems in data science, but are notoriously difficult to train. These difficulties come from the fact that optimal weights for adversarial nets correspond to saddle points, and not minimizers, of the loss function. The alternating stochastic gradient methods typically used for such problems do not reliably converge to saddle points, and when convergence does happen it is often highly sensitive to learning rates. We propose a simple modification of stochastic gradient descent that stabilizes adversarial networks. We show, both in theory and practice, that the proposed method reliably converges to saddle points. This makes adversarial networks less likely to "collapse," and enables faster training with larger learning rates.
TL;DR:We present a simple modification to the alternating SGD method, called a prediction step, that improves the stability of adversarial networks.
Keywords:adversarial networks, optimization
Enter your feedback below and we'll get back to you as soon as possible.