Clustering-friendly Representation Learning for Enhancing Salient Features

TL;DR

This paper introduces a novel unsupervised representation learning method that emphasizes features important for image clustering, improving clustering performance over existing contrastive and deep clustering techniques.

Contribution

It extends contrastive learning with a clustering-friendly approach and a contrastive analysis method that uses a reference dataset to distinguish important features.

Findings

Higher clustering scores across three datasets

Outperforms conventional contrastive analysis methods

Enhances feature importance for clustering tasks

Abstract

Recently, representation learning with contrastive learning algorithms has been successfully applied to challenging unlabeled datasets. However, these methods are unable to distinguish important features from unimportant ones under simply unsupervised settings, and definitions of importance vary according to the type of downstream task or analysis goal, such as the identification of objects or backgrounds. In this paper, we focus on unsupervised image clustering as the downstream task and propose a representation learning method that enhances features critical to the clustering task. We extend a clustering-friendly contrastive learning method and incorporate a contrastive analysis approach, which utilizes a reference dataset to separate important features from unimportant ones, into the design of loss functions. Conducting an experimental evaluation of image clustering for three datasets with characteristic backgrounds, we show that for all datasets, our method achieves higher clustering scores compared with conventional contrastive analysis and deep clustering methods.