Generative Adversarial Networks with Inverse Transformation Unit

2017/09/27 by Zhifeng Kong, Kong, Zhifeng, Shuo Ding +1
Computer Science · #Advanced Steganography and Watermarking Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Generative Adversarial Networks and Image Synthesis #Image Processing and 3D Reconstruction #cs.CV

paper · pdf · doi:10.48550/arxiv.1709.09354

11 pages, 9 figures

arxiv created 2017/09/27 · openalex publication_date 2017/09/27 · arxiv updated 2017/09/28 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

In this paper we introduce a new structure to Generative Adversarial Networks by adding an inverse transformation unit behind the generator. We present two theorems to claim the convergence of the model, and two conjectures to nonideal situations when the transformation is not bijection. A general survey on models with different transformations was done on the MNIST dataset and the Fashion-MNIST dataset, which shows the transformation does not necessarily need to be bijection. Also, with certain transformations that blurs an image, our model successfully learned to sharpen the images and recover blurred images, which was additionally verified by our measurement of sharpness.

Citations

• Gradient-based learning applied to document recognition
• From A to Z: Supervised Transfer of Style and Content Using Deep Neural Network Generators
• Learning Typographic Style
• f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization
• Energy-based Generative Adversarial Network
• Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets

Related

• Autoencoding Generative Adversarial Networks
• Generative Adversarial Networks: An Overview
• An Empirical Study of Generative Models with Encoders
• IVE-GAN: Invariant Encoding Generative Adversarial Networks
• Learning Inverse Mapping by Autoencoder based Generative Adversarial Nets