Bootstrap Deep Spectral Clustering with Optimal Transport

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Guo, Wengang, Ye, Wei, Chen, Chunchun, Sun, Xin, Böhm, Christian, Plant, Claudia, Rahardja, Susanto
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915430739738624
author Guo, Wengang
Ye, Wei
Chen, Chunchun
Sun, Xin
Böhm, Christian
Plant, Claudia
Rahardja, Susanto
author_facet Guo, Wengang
Ye, Wei
Chen, Chunchun
Sun, Xin
Böhm, Christian
Plant, Claudia
Rahardja, Susanto
contents Spectral clustering is a leading clustering method. Two of its major shortcomings are the disjoint optimization process and the limited representation capacity. To address these issues, we propose a deep spectral clustering model (named BootSC), which jointly learns all stages of spectral clustering -- affinity matrix construction, spectral embedding, and $k$-means clustering -- using a single network in an end-to-end manner. BootSC leverages effective and efficient optimal-transport-derived supervision to bootstrap the affinity matrix and the cluster assignment matrix. Moreover, a semantically-consistent orthogonal re-parameterization technique is introduced to orthogonalize spectral embeddings, significantly enhancing the discrimination capability. Experimental results indicate that BootSC achieves state-of-the-art clustering performance. For example, it accomplishes a notable 16\% NMI improvement over the runner-up method on the challenging ImageNet-Dogs dataset. Our code is available at https://github.com/spdj2271/BootSC.
format Preprint
id arxiv_https___arxiv_org_abs_2508_04200
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Bootstrap Deep Spectral Clustering with Optimal Transport
Guo, Wengang
Ye, Wei
Chen, Chunchun
Sun, Xin
Böhm, Christian
Plant, Claudia
Rahardja, Susanto
Computer Vision and Pattern Recognition
Machine Learning
Spectral clustering is a leading clustering method. Two of its major shortcomings are the disjoint optimization process and the limited representation capacity. To address these issues, we propose a deep spectral clustering model (named BootSC), which jointly learns all stages of spectral clustering -- affinity matrix construction, spectral embedding, and $k$-means clustering -- using a single network in an end-to-end manner. BootSC leverages effective and efficient optimal-transport-derived supervision to bootstrap the affinity matrix and the cluster assignment matrix. Moreover, a semantically-consistent orthogonal re-parameterization technique is introduced to orthogonalize spectral embeddings, significantly enhancing the discrimination capability. Experimental results indicate that BootSC achieves state-of-the-art clustering performance. For example, it accomplishes a notable 16\% NMI improvement over the runner-up method on the challenging ImageNet-Dogs dataset. Our code is available at https://github.com/spdj2271/BootSC.
title Bootstrap Deep Spectral Clustering with Optimal Transport
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2508.04200