Generalized Dual Discriminator GANs

TL;DR

This paper introduces a generalized class of dual discriminator GANs that combine flexible loss functions with theoretical guarantees, effectively addressing mode collapse and improving generative performance.

Contribution

It extends dual discriminator GANs with a tunable loss and broad theoretical analysis, unifying various divergence measures within a single framework.

Findings

The proposed models reduce to a combination of $f$-divergences and reverse $f$-divergences.

Experimental results demonstrate advantages on synthetic data.

Theoretical analysis confirms the minimization of a linear combination of divergences.

Abstract

Dual discriminator generative adversarial networks (D2 GANs) were introduced to mitigate the problem of mode collapse in generative adversarial networks. In D2 GANs, two discriminators are employed alongside a generator: one discriminator rewards high scores for samples from the true data distribution, while the other favors samples from the generator. In this work, we first introduce dual discriminator $\alpha$-GANs (D2 $\alpha$-GANs), which combines the strengths of dual discriminators with the flexibility of a tunable loss function, $\alpha$-loss. We further generalize this approach to arbitrary functions defined on positive reals, leading to a broader class of models we refer to as generalized dual discriminator generative adversarial networks. For each of these proposed models, we provide theoretical analysis and show that the associated min-max optimization reduces to the minimization of a linear combination of an $f$-divergence and a reverse $f$-divergence. This generalizes the known simplification for D2-GANs, where the objective reduces to a linear combination of the KL-divergence and the reverse KL-divergence. Finally, we perform experiments on 2D synthetic data and use multiple performance metrics to capture various advantages of our GANs.