DO-GAN: A Double Oracle Framework for Generative Adversarial Networks

DO-GAN: A Double Oracle Framework for Generative Adversarial Networks

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

10.1109/CVPR52688.2022.01099

http://dx.doi.org/10.1109/cvpr52688.2022.01099

Aye Phyu Phyu Aung, Xinrun Wang, Runsheng Yu, Bo An, Senthilnath Jayavelu, Xiaoli Li

27 September 2022

Phyu Aung, A. P., Wang, X., Yu, R., An, B., Jayavelu, S., & Li, X. (2022). DO-GAN: A Double Oracle Framework for Generative Adversarial Networks. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). https://doi.org/10.1109/cvpr52688.2022.01099

In this paper, we propose a new approach to train Generative Adversarial Networks (GANs) where we deploy a double-oracle framework using the generator and discriminator oracles. GAN is essentially a two-player zero-sum game between the generator and the discriminator. Training GANs is challenging as a pure Nash equilibrium may not exist and even finding the mixed Nash equilibrium is difficult as GANs have a large-scale strategy space. In DOGAN, we extend the double oracle framework to GANs. We first generalize the players’ strategies as the trained models of generator and discriminator from the best response oracles. We then compute the meta-strategies using a linear program. For scalability of the framework where multiple generators and discriminator best responses are stored in the memory, we propose two solutions: 1) pruning the weakly-dominated players’ strategies to keep the oracles from becoming intractable; 2) applying continual learning to retain the previous knowledge of the networks. We apply our framework to established GAN architectures such as vanilla GAN, Deep Convolutional GAN, Spectral Normalization GAN and Stacked GAN. Finally, we conduct experiments on MNIST, CIFAR-10 and CelebA datasets and show that DO-GAN variants have significant improvements in both subjective qualitative evaluation and quantitative metrics, compared with their respective GAN architectures.

Publisher Copyright

This research / project is supported by the National Research Foundation - AI Singapore Programme
Grant Reference no. : AISG-RP-2019-0013

This research / project is supported by the National Satellite of Excellence in Trustworthy Software Systems - Trustworthy Software Systems – Core Technologies
Grant Reference no. : NSOE-TSS2019-01

© 2022 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

https://oar.a-star.edu.sg/communities-collections/articles/18677

NA

Institute for Infocomm Research