Distributional Adversarial Networks
This addresses a key bottleneck in generative adversarial networks for researchers and practitioners by enhancing training stability and reducing mode collapse.
The paper tackles the instability and mode collapse in adversarial training by introducing distributional adversaries that operate on samples rather than single points, resulting in more stable generators and significant improvements in domain adaptation over state-of-the-art methods.
We propose a framework for adversarial training that relies on a sample rather than a single sample point as the fundamental unit of discrimination. Inspired by discrepancy measures and two-sample tests between probability distributions, we propose two such distributional adversaries that operate and predict on samples, and show how they can be easily implemented on top of existing models. Various experimental results show that generators trained with our distributional adversaries are much more stable and are remarkably less prone to mode collapse than traditional models trained with pointwise prediction discriminators. The application of our framework to domain adaptation also results in considerable improvement over recent state-of-the-art.