Automatic Data Curation for Self-Supervised Learning: A Clustering-Based Approach
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917707984666624 |
|---|---|
| author | Vo, Huy V. Khalidov, Vasil Darcet, Timothée Moutakanni, Théo Smetanin, Nikita Szafraniec, Marc Touvron, Hugo Couprie, Camille Oquab, Maxime Joulin, Armand Jégou, Hervé Labatut, Patrick Bojanowski, Piotr |
| author_facet | Vo, Huy V. Khalidov, Vasil Darcet, Timothée Moutakanni, Théo Smetanin, Nikita Szafraniec, Marc Touvron, Hugo Couprie, Camille Oquab, Maxime Joulin, Armand Jégou, Hervé Labatut, Patrick Bojanowski, Piotr |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_15613 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Automatic Data Curation for Self-Supervised Learning: A Clustering-Based Approach Vo, Huy V. Khalidov, Vasil Darcet, Timothée Moutakanni, Théo Smetanin, Nikita Szafraniec, Marc Touvron, Hugo Couprie, Camille Oquab, Maxime Joulin, Armand Jégou, Hervé Labatut, Patrick Bojanowski, Piotr Machine Learning Artificial Intelligence Computer Vision and Pattern Recognition 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. |
| title | Automatic Data Curation for Self-Supervised Learning: A Clustering-Based Approach |
| topic | Machine Learning Artificial Intelligence Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2405.15613 |