Multi-Modal Deep Clustering: Unsupervised Partitioning of Images

TL;DR

This paper introduces Multi-Modal Deep Clustering (MMDC), an end-to-end unsupervised deep learning framework that clusters images by aligning embeddings with a Gaussian Mixture Model and improves representation learning through self-supervised rotation prediction.

Contribution

The paper presents a novel unsupervised clustering method that combines deep neural network training with Gaussian Mixture Model alignment and self-supervised rotation prediction for improved image clustering.

Findings

Achieves or exceeds state-of-the-art results on six benchmarks.

Improves clustering accuracy by up to 20% on natural image datasets.

Attains 82% accuracy on CIFAR-10, 45% on CIFAR-100, and 69% on STL-10.

Abstract

The clustering of unlabeled raw images is a daunting task, which has recently been approached with some success by deep learning methods. Here we propose an unsupervised clustering framework, which learns a deep neural network in an end-to-end fashion, providing direct cluster assignments of images without additional processing. Multi-Modal Deep Clustering (MMDC), trains a deep network to align its image embeddings with target points sampled from a Gaussian Mixture Model distribution. The cluster assignments are then determined by mixture component association of image embeddings. Simultaneously, the same deep network is trained to solve an additional self-supervised task of predicting image rotations. This pushes the network to learn more meaningful image representations that facilitate a better clustering. Experimental results show that MMDC achieves or exceeds state-of-the-art performance on six challenging benchmarks. On natural image datasets we improve on previous results with significant margins of up to 20% absolute accuracy points, yielding an accuracy of 82% on CIFAR-10, 45% on CIFAR-100 and 69% on STL-10.