Precise Recovery of Latent Vectors from Generative Adversarial Networks

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.

Cited by in corpus (15)

• Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis
• Unsupervised Image-to-Image Translation via Pre-trained StyleGAN2 Network
• Semantic Segmentation with Generative Models: Semi-Supervised Learning and Strong Out-of-Domain Generalization
• Learning a low dimensional manifold of real cancer tissue with PathologyGAN
• Learning Formation of Physically-Based Face Attributes
• Latent Variable Modeling for Generative Concept Representations and Deep Generative Models
• Deep Meditations: Controlled navigation of latent space
• GAN-Based Facial Attractiveness Enhancement
• Exploiting GAN Internal Capacity for High-Quality Reconstruction of Natural Images
• Inverse mapping of face GANs
• Neural Crossbreed: Neural Based Image Metamorphosis
• Matched sample selection with GANs for mitigating attribute confounding
• Ensembles of Generative Adversarial Networks for Disconnected Data
• Force-in-domain GAN inversion
• The Effects of Invertibility on the Representational Complexity of Encoders in Variational Autoencoders