Optimizing the Latent Space of Generative Networks
This work addresses the challenge of simplifying GAN training for researchers and practitioners by removing adversarial optimization, though it is incremental as it builds on existing GAN components.
The paper tackled the problem of disentangling the contributions of adversarial optimization and deep convolutional networks in GANs by introducing Generative Latent Optimization (GLO), a framework that trains deep convolutional generators using simple reconstruction losses, achieving visually-appealing samples, meaningful interpolation, and linear arithmetic without adversarial schemes.
Generative Adversarial Networks (GANs) have achieved remarkable results in the task of generating realistic natural images. In most successful applications, GAN models share two common aspects: solving a challenging saddle point optimization problem, interpreted as an adversarial game between a generator and a discriminator functions; and parameterizing the generator and the discriminator as deep convolutional neural networks. The goal of this paper is to disentangle the contribution of these two factors to the success of GANs. In particular, we introduce Generative Latent Optimization (GLO), a framework to train deep convolutional generators using simple reconstruction losses. Throughout a variety of experiments, we show that GLO enjoys many of the desirable properties of GANs: synthesizing visually-appealing samples, interpolating meaningfully between samples, and performing linear arithmetic with noise vectors; all of this without the adversarial optimization scheme.