CoDeGAN: Contrastive Disentanglement for Generative Adversarial Network

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Jiangwei, Liu, Zejia, Guo, Xiaohan, Pan, Lili
Format: Preprint
Published: 2021
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913371487469568
author Zhao, Jiangwei
Liu, Zejia
Guo, Xiaohan
Pan, Lili
author_facet Zhao, Jiangwei
Liu, Zejia
Guo, Xiaohan
Pan, Lili
contents Disentanglement, a critical concern in interpretable machine learning, has also garnered significant attention from the computer vision community. Many existing GAN-based class disentanglement (unsupervised) approaches, such as InfoGAN and its variants, primarily aim to maximize the mutual information (MI) between the generated image and its latent codes. However, this focus may lead to a tendency for the network to generate highly similar images when presented with the same latent class factor, potentially resulting in mode collapse or mode dropping. To alleviate this problem, we propose \texttt{CoDeGAN} (Contrastive Disentanglement for Generative Adversarial Networks), where we relax similarity constraints for disentanglement from the image domain to the feature domain. This modification not only enhances the stability of GAN training but also improves their disentangling capabilities. Moreover, we integrate self-supervised pre-training into CoDeGAN to learn semantic representations, significantly facilitating unsupervised disentanglement. Extensive experimental results demonstrate the superiority of our method over state-of-the-art approaches across multiple benchmarks. The code is available at https://github.com/learninginvision/CoDeGAN.
format Preprint
id arxiv_https___arxiv_org_abs_2103_03636
institution arXiv
publishDate 2021
record_format arxiv
spellingShingle CoDeGAN: Contrastive Disentanglement for Generative Adversarial Network
Zhao, Jiangwei
Liu, Zejia
Guo, Xiaohan
Pan, Lili
Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
Disentanglement, a critical concern in interpretable machine learning, has also garnered significant attention from the computer vision community. Many existing GAN-based class disentanglement (unsupervised) approaches, such as InfoGAN and its variants, primarily aim to maximize the mutual information (MI) between the generated image and its latent codes. However, this focus may lead to a tendency for the network to generate highly similar images when presented with the same latent class factor, potentially resulting in mode collapse or mode dropping. To alleviate this problem, we propose \texttt{CoDeGAN} (Contrastive Disentanglement for Generative Adversarial Networks), where we relax similarity constraints for disentanglement from the image domain to the feature domain. This modification not only enhances the stability of GAN training but also improves their disentangling capabilities. Moreover, we integrate self-supervised pre-training into CoDeGAN to learn semantic representations, significantly facilitating unsupervised disentanglement. Extensive experimental results demonstrate the superiority of our method over state-of-the-art approaches across multiple benchmarks. The code is available at https://github.com/learninginvision/CoDeGAN.
title CoDeGAN: Contrastive Disentanglement for Generative Adversarial Network
topic Computer Vision and Pattern Recognition
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2103.03636