Generalized Latent Variable Recovery for Generative Adversarial Networks
This work addresses a specific technical challenge in GANs for researchers and practitioners, but it is incremental as it builds on existing methods.
The paper tackled the problem of projecting input images onto the latent space of Generative Adversarial Networks (GANs) with Gaussian priors, extending previous techniques from uniform priors and demonstrating effectiveness.
The Generator of a Generative Adversarial Network (GAN) is trained to transform latent vectors drawn from a prior distribution into realistic looking photos. These latent vectors have been shown to encode information about the content of their corresponding images. Projecting input images onto the latent space of a GAN is non-trivial, but previous work has successfully performed this task for latent spaces with a uniform prior. We extend these techniques to latent spaces with a Gaussian prior, and demonstrate our technique's effectiveness.