Deep Clustering for Unsupervised Learning of Visual Features

TL;DR

DeepCluster introduces an unsupervised learning method that jointly trains neural network features and cluster assignments, significantly improving performance on large-scale image datasets without labeled data.

Contribution

It presents DeepCluster, a novel end-to-end clustering approach that combines neural network training with iterative k-means clustering for unsupervised visual feature learning.

Findings

Outperforms state-of-the-art methods on ImageNet and YFCC100M benchmarks.

Effectively learns visual features without labeled data.

Demonstrates scalability to large datasets.

Abstract

Clustering is a class of unsupervised learning methods that has been extensively applied and studied in computer vision. Little work has been done to adapt it to the end-to-end training of visual features on large scale datasets. In this work, we present DeepCluster, a clustering method that jointly learns the parameters of a neural network and the cluster assignments of the resulting features. DeepCluster iteratively groups the features with a standard clustering algorithm, k-means, and uses the subsequent assignments as supervision to update the weights of the network. We apply DeepCluster to the unsupervised training of convolutional neural networks on large datasets like ImageNet and YFCC100M. The resulting model outperforms the current state of the art by a significant margin on all the standard benchmarks.