ComGAN: Toward GANs Exploiting Multiple Samples

Haeone Lee
4/24/2023
cs.LGcs.CV

No official code found. Best implementation: rl-max/ComGAN (3 stars, last commit 2023-04-26).

Abstract

In this paper, we propose ComGAN(ComparativeGAN) which allows the generator in GANs to refer to the semantics of comparative samples(e.g. real data) by comparison. ComGAN generalizes relativistic GANs by using arbitrary architecture and mostly outperforms relativistic GANs in simple input-concatenation architecture. To train the discriminator in ComGAN, we also propose equality regularization, which fits the discriminator to a neutral label for equally real or fake samples. Equality regularization highly boosts the performance of ComGAN including WGAN while being exceptionally simple compared to existing regularizations. Finally, we generalize comparative samples fixed to real data in relativistic GANs toward fake data and show that such objectives are sound in both theory and practice. Our experiments demonstrate superior performances of ComGAN and equality regularization, achieving the best FIDs in 7 out of 8 cases of different losses and data against ordinary GANs and relativistic GANs.

View on arXivView PDF

Code Implementations(1)

rl-max/ComGANCommunity100%

StudioGAN is a Pytorch library providing implementations of representative Generative Adversarial Networks (GANs) for conditional/unconditional image generation.

30C++, CudaOct 30, 20223 years agoNOASSERTION

Similar Papers

The GAN is dead; long live the GAN! A Modern GAN Baseline

79% match1 repo8762025

The relativistic discriminator: a key element missing from standard GAN

78% match1 repo7422018

MS$^3$D: A RG Flow-Based Regularization for GAN Training with Limited Data

75% match2024

Cross-GAN Auditing: Unsupervised Identification of Attribute Level Similarities and Differences between Pretrained Generative Models

75% match2023

Imbalance-Robust and Sampling-Efficient Continuous Conditional GANs via Adaptive Vicinity and Auxiliary Regularization

75% match1 repo12025

Improving SCGAN's Similarity Constraint and Learning a Better Disentangled Representation

75% match2023

Cite this paper

@article{lee2023comgan,
  title  = {ComGAN: Toward GANs Exploiting Multiple Samples},
  author = {Haeone Lee},
  year   = {2023},
  eprint = {2304.12098},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2304.12098v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2304.12098.svg)](https://codeofpaper.com/paper/2304.12098)

Discussion