An Empirical Study into Clustering of Unseen Datasets with Self-Supervised Encoders

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Lowe, Scott C., Haurum, Joakim Bruslund, Oore, Sageev, Moeslund, Thomas B., Taylor, Graham W.
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916273912283136
author Lowe, Scott C.
Haurum, Joakim Bruslund
Oore, Sageev
Moeslund, Thomas B.
Taylor, Graham W.
author_facet Lowe, Scott C.
Haurum, Joakim Bruslund
Oore, Sageev
Moeslund, Thomas B.
Taylor, Graham W.
contents Can pretrained models generalize to new datasets without any retraining? We deploy pretrained image models on datasets they were not trained for, and investigate whether their embeddings form meaningful clusters. Our suite of benchmarking experiments use encoders pretrained solely on ImageNet-1k with either supervised or self-supervised training techniques, deployed on image datasets that were not seen during training, and clustered with conventional clustering algorithms. This evaluation provides new insights into the embeddings of self-supervised models, which prioritize different features to supervised models. Supervised encoders typically offer more utility than SSL encoders within the training domain, and vice-versa far outside of it, however, fine-tuned encoders demonstrate the opposite trend. Clustering provides a way to evaluate the utility of self-supervised learned representations orthogonal to existing methods such as kNN. Additionally, we find the silhouette score when measured in a UMAP-reduced space is highly correlated with clustering performance, and can therefore be used as a proxy for clustering performance on data with no ground truth labels. Our code implementation is available at \url{https://github.com/scottclowe/zs-ssl-clustering/}.
format Preprint
id arxiv_https___arxiv_org_abs_2406_02465
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle An Empirical Study into Clustering of Unseen Datasets with Self-Supervised Encoders
Lowe, Scott C.
Haurum, Joakim Bruslund
Oore, Sageev
Moeslund, Thomas B.
Taylor, Graham W.
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Can pretrained models generalize to new datasets without any retraining? We deploy pretrained image models on datasets they were not trained for, and investigate whether their embeddings form meaningful clusters. Our suite of benchmarking experiments use encoders pretrained solely on ImageNet-1k with either supervised or self-supervised training techniques, deployed on image datasets that were not seen during training, and clustered with conventional clustering algorithms. This evaluation provides new insights into the embeddings of self-supervised models, which prioritize different features to supervised models. Supervised encoders typically offer more utility than SSL encoders within the training domain, and vice-versa far outside of it, however, fine-tuned encoders demonstrate the opposite trend. Clustering provides a way to evaluate the utility of self-supervised learned representations orthogonal to existing methods such as kNN. Additionally, we find the silhouette score when measured in a UMAP-reduced space is highly correlated with clustering performance, and can therefore be used as a proxy for clustering performance on data with no ground truth labels. Our code implementation is available at \url{https://github.com/scottclowe/zs-ssl-clustering/}.
title An Empirical Study into Clustering of Unseen Datasets with Self-Supervised Encoders
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2406.02465