Precise Recovery of Latent Vectors from Generative Adversarial Networks

TL;DR

This paper introduces a gradient-based method called stochastic clipping that accurately recovers latent vectors from GAN-generated images, achieving perfect recovery and robustness to noise, thus enabling effective inversion of GANs.

Contribution

The paper presents a simple, effective technique for precisely recovering latent vectors from GAN images, addressing the challenge of GAN inversion.

Findings

Achieves 100% recovery of latent vectors for generated images

Demonstrates robustness of the method to noise

Recovers unique encodings for unseen images

Abstract

Generative adversarial networks (GANs) transform latent vectors into visually plausible images. It is generally thought that the original GAN formulation gives no out-of-the-box method to reverse the mapping, projecting images back into latent space. We introduce a simple, gradient-based technique called stochastic clipping. In experiments, for images generated by the GAN, we precisely recover their latent vector pre-images 100% of the time. Additional experiments demonstrate that this method is robust to noise. Finally, we show that even for unseen images, our method appears to recover unique encodings.