Automatic Data Curation for Self-Supervised Learning: A Clustering-Based Approach

TL;DR

This paper introduces a clustering-based method for automatically curating high-quality, diverse, and balanced datasets for self-supervised learning, reducing manual effort and improving feature quality across multiple data domains.

Contribution

The authors propose a novel hierarchical clustering and sampling approach for automatic dataset curation tailored for self-supervised pre-training, demonstrating its effectiveness across various data types.

Findings

Features trained on curated datasets outperform uncurated data.

Automatically curated datasets match or surpass manually curated datasets.

Method is effective across images and text domains.

Abstract

Self-supervised features are the cornerstone of modern machine learning systems. They are typically pre-trained on data collections whose construction and curation typically require extensive human effort. This manual process has some limitations similar to those encountered in supervised learning, e.g., the crowd-sourced selection of data is costly and time-consuming, preventing scaling the dataset size. In this work, we consider the problem of automatic curation of high-quality datasets for self-supervised pre-training. We posit that such datasets should be large, diverse and balanced, and propose a clustering-based approach for building ones satisfying all these criteria. Our method involves successive and hierarchical applications of $k$-means on a large and diverse data repository to obtain clusters that distribute uniformly among data concepts, followed by a hierarchical, balanced sampling step from these clusters. Extensive experiments on three different data domains including web-based images, satellite images and text show that features trained on our automatically curated datasets outperform those trained on uncurated data while being on par or better than ones trained on manually curated data. Code is available at https://github.com/facebookresearch/ssl-data-curation.