Joint Debiased Representation Learning and Imbalanced Data Clustering

TL;DR

This paper introduces a novel unsupervised framework that jointly learns debiased representations and performs image clustering, effectively handling class imbalance and out-of-distribution samples to improve clustering accuracy and generalization.

Contribution

The proposed method combines deep representation learning with clustering using a statistics pooling block to address class imbalance and out-of-distribution issues in unsupervised image clustering.

Findings

Significantly improves clustering results on imbalanced datasets

Enhances generalization to out-of-distribution datasets

Utilizes a novel statistics pooling block for debiased learning

Abstract

One of the most promising approaches for unsupervised learning is combining deep representation learning and deep clustering. Some recent works propose to simultaneously learn representation using deep neural networks and perform clustering by defining a clustering loss on top of embedded features. However, these approaches are sensitive to imbalanced data and out-of-distribution samples. As a consequence, these methods optimize clustering by pushing data close to randomly initialized cluster centers. This is problematic when the number of instances varies largely in different classes or a cluster with few samples has less chance to be assigned a good centroid. To overcome these limitations, we introduce a new unsupervised framework for joint debiased representation learning and image clustering. We simultaneously train two deep learning models, a deep representation network that captures the data distribution, and a deep clustering network that learns embedded features and performs clustering. Specifically, the clustering network and learning representation network both take advantage of our proposed statistics pooling block that represents mean, variance, and cardinality to handle the out-of-distribution samples and class imbalance. Our experiments show that using these representations, one can considerably improve results on imbalanced image clustering across a variety of image datasets. Moreover, the learned representations generalize well when transferred to the out-of-distribution dataset.