Learning Discrete Structured Representations by Adversarially Maximizing Mutual Information

Open in new window