Abstract:Modern unsupervised feature selection methods predominantly obtain the cluster structure and pseudo -labels information through spectral clustering. However, the pseudo-labels obtained by spectral clustering are usually mixed between positive and negative. Moreover, the Laplacian matrix in spectral clustering typically affects f eature selection. Additionally, spectral clustering does not consider the interconnection information between data. To address these problems, this paper proposes uncorrelated feature selec-tion via sparse latent representation and extended orthogonal least square discriminant analysis (OLSDA), which we term SLREO). Firstly, SLREO retains the interconnection between data by latent representa-tion learning, and preserves the internal information between the data. In order to remove redundant interconnection information, an l 2,1-norm constraint is applied to the residual matrix of potential repre-sentation learning. Secondly, SLREO obtains non-negative pseudo-labels through orthogonal least square discriminant analysis (OLSDA) of embedded non-negative manifold structure. It not only avoids the ap-pearance of negative pseudo-labels, but also eliminates the effect of the Laplacian matrix on feature se-lection. The manifold information of the data is also preserved. Furthermore, the matrix of the learned latent representation and OLSDA is used as pseudo-labels information. It not only ensures that the gen-erated pseudo-labels are non-negative, but also makes the pseudo-labels closer to the true class labels. Finally, in order to avoid trivial solutions, an uncorrelated constraint and l 2,1-norm constraint are imposed on the feature transformation matrix. These constraints ensure row sparsity of the feature transformation matrix, select low-redundant and discriminative features, and improve the effect of feature selection. Ex-perimental results show that the Clustering Accuracy (ACC) and Normalized Mutual Information (NMI) of SLREO are significantly improved, as compared with six other published algorithms, tested on 11 bench-mark datasets.(c) 2022 Elsevier Ltd. All rights reserved.

Feature selection based on non-negative spectral feature learning and adaptive rank constraint

Unsupervised Feature Selection Using Nonnegative Spectral Analysis.

Unsupervised Feature Selection by Nonnegative Sparsity Adaptive Subspace Learning

Nonnegative Spectral Analysis with Adaptive Graph and L 2, 0 -Norm Regularization for Unsupervised Feature Selection.

Graph-regularized Feature Selection Based on Spectral Learning and Subspace Learning

Adaptive graph regularization and self-expression for noise-aware feature selection

Nonnegative Spectral Analysis with Adaptive Graph and $L_{2,0}$-Norm Regularization for Unsupervised Feature Selection

Unsupervised Feature Selection via Nonnegative Orthogonal Constrained Regularized Minimization

Attention Non-Negative Spectral Clustering

Clustering-Guided Sparse Structural Learning for Unsupervised Feature Selection

Discriminative and Uncorrelated Feature Selection with Constrained Spectral Analysis in Unsupervised Learning

Convex Non-Negative Matrix Factorization with Adaptive Graph for Unsupervised Feature Selection.

Double feature selection algorithm based on low-rank sparse non-negative matrix factorization

Uncorrelated Feature Selection Via Sparse Latent Representation and Extended OLSDA

Adaptive Dual Graphs and Non-Convex Constraint Based Embedded Feature Selection

Unsupervised Feature Selection Via Neural Networks and Self-Expression with Adaptive Graph Constraint

Robust Unsupervised Feature Selection by Nonnegative Sparse Subspace Learning.

Unsupervised Feature Selection with Graph Learning Via Low-Rank Constraint

Unsupervised feature selection via discrete spectral clustering and feature weights

Sparse Feature Selection Via Fast Embedding Spectral Analysis

Feature Selection Based Dual-Graph Sparse Non-Negative Matrix Factorization for Local Discriminative Clustering.