On Matrix Factorizations in Subspace Clustering

TL;DR

This paper investigates the use of CUR matrix decompositions in subspace clustering, analyzing how different hyperparameters affect performance on real-world datasets and providing practical guidelines.

Contribution

It introduces the application of CUR decompositions to subspace clustering and systematically studies hyperparameter effects with experimental guidelines.

Findings

Hyperparameters significantly influence clustering accuracy.

Sampling methods impact the effectiveness of CUR-based clustering.

Practical parameter guidelines are proposed for real-world applications.

Abstract

This article explores subspace clustering algorithms using CUR decompositions, and examines the effect of various hyperparameters in these algorithms on clustering performance on two real-world benchmark datasets, the Hopkins155 motion segmentation dataset and the Yale face dataset. Extensive experiments are done for a variety of sampling methods and oversampling parameters for these datasets, and some guidelines for parameter choices are given for practical applications.