# Clustering Images by Unmasking - A New Baseline

## TL;DR

This paper introduces a novel image clustering method based on unmasking, which iteratively trains classifiers and removes discriminant features to determine cluster similarity, demonstrating improved performance across multiple tasks.

## Contribution

It is the first to apply unmasking for image clustering, offering a new baseline that outperforms traditional methods like k-means and recent state-of-the-art approaches.

## Key findings

- Improves clustering performance across various tasks
- Effective with both deep and shallow features
- Outperforms k-means and recent methods

## Abstract

We propose a novel agglomerative clustering method based on unmasking, a technique that was previously used for authorship verification of text documents and for abnormal event detection in videos. In order to join two clusters, we alternate between (i) training a binary classifier to distinguish between the samples from one cluster and the samples from the other cluster, and (ii) removing at each step the most discriminant features. The faster-decreasing accuracy rates of the intermediately-obtained classifiers indicate that the two clusters should be joined. To the best of our knowledge, this is the first work to apply unmasking in order to cluster images. We compare our method with k-means as well as a recent state-of-the-art clustering method. The empirical results indicate that our approach is able to improve performance for various (deep and shallow) feature representations and different tasks, such as handwritten digit recognition, texture classification and fine-grained object recognition.