Maximum-Likelihood Augmented Discrete Generative Adversarial Networks
Abstract
Despite the successes in capturing continuous distributions, the application of generative adversarial networks (GANs) to discrete settings, like natural language tasks, is rather restricted. The fundamental reason is the difficulty of back-propagation through discrete random variables combined with the inherent instability of the GAN training objective. To address these problems, we propose Maximum-Likelihood Augmented Discrete Generative Adversarial Networks. Instead of directly optimizing the GAN objective, we derive a novel and low-variance objective using the discriminator's output that follows corresponds to the log-likelihood. Compared with the original, the new objective is proved to be consistent in theory and beneficial in practice. The experimental results on various discrete datasets demonstrate the effectiveness of the proposed approach.
Cite
@article{arxiv.1702.07983,
title = {Maximum-Likelihood Augmented Discrete Generative Adversarial Networks},
author = {Tong Che and Yanran Li and Ruixiang Zhang and R Devon Hjelm and Wenjie Li and Yangqiu Song and Yoshua Bengio},
journal= {arXiv preprint arXiv:1702.07983},
year = {2017}
}
Comments
11 pages, 3 figures