Clustering with feature selection using alternating minimization. Application to computational biology

Abstract : This paper deals with unsupervised clustering with feature selection. The problem is to estimate both labels and a sparse projection matrix of weights. To address this combina-torial non-convex problem maintaining a strict control on the sparsity of the matrix of weights, we propose an alternating minimization of the Frobenius norm criterion. We provide a new efficient algorithm named K-sparse which alternates k-means with projection-gradient minimization. The projection-gradient step is a method of splitting type, with exact projection on the ℓ 1 ball to promote sparsity. The convergence of the gradient-projection step is addressed, and a preliminary analysis of the alternating minimization is made. The Frobenius norm criterion converges as the number of iterates in Algorithm K-sparse goes to infinity. Experiments on Single Cell RNA sequencing datasets show that our method significantly improves the results of PCA k-means, spectral clustering, SIMLR, and Sparcl methods. The complexity of K-sparse is linear in the number of samples (cells), so that the method scales up to large datasets. Finally, we extend K-sparse to supervised classification.
Document type :
Preprints, Working Papers, ...
Complete list of metadatas

Cited literature [42 references]  Display  Hide  Download

https://hal.inria.fr/hal-01671982
Contributor : Jean-Baptiste Caillau <>
Submitted on : Friday, December 22, 2017 - 6:02:16 PM
Last modification on : Friday, January 25, 2019 - 10:00:06 AM

File

cluster.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01671982, version 1

Citation

Cyprien Gilet, Marie Deprez, Jean-Baptiste Caillau, Michel Barlaud. Clustering with feature selection using alternating minimization. Application to computational biology. 2017. ⟨hal-01671982⟩

Share

Metrics

Record views

531

Files downloads

105