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

TL;DR

This paper enhances SCGAN by replacing the similarity constraint with SSIM and contrastive loss principles, leading to improved disentangled representations and better generalization as evidenced by FID and FactorVAE metrics.

Contribution

The paper introduces a modified SCGAN that incorporates SSIM and contrastive loss, improving disentanglement and generalization over the original model.

Findings

Improved FID and FactorVAE scores with the modified model.

Enhanced generalizability compared to baseline models.

Better disentangled representations achieved.

Abstract

SCGAN adds a similarity constraint between generated images and conditions as a regularization term on generative adversarial networks. Similarity constraint works as a tutor to instruct the generator network to comprehend the difference of representations based on conditions. We understand how SCGAN works on a deeper level. This understanding makes us realize that the similarity constraint functions like the contrastive loss function. We believe that a model with high understanding and intelligence measures the similarity between images based on their structure and high level features, just like humans do. Two major changes we applied to SCGAN in order to make a modified model are using SSIM to measure similarity between images and applying contrastive loss principles to the similarity constraint. The modified model performs better using FID and FactorVAE metrics. The modified model also has better generalisability compared to other models. Keywords Generative Adversarial Nets, Unsupervised Learning, Disentangled Representation Learning, Contrastive Disentanglement, SSIM