A ‎robust ‎‎unsupervised ‎‎feature ‎s‎election based on ‎‎subspace ‎l‎earning and ‎‎adaptive ‎‎graph ‎‎structure

Sohrabi, Hazhir; Esmaeili, Shahrokh; Moradi, Parham

doi:10.22124/jmm.2025.30225.2714

A ‎robust ‎‎unsupervised ‎‎feature ‎s‎election based on ‎‎subspace ‎l‎earning and ‎‎adaptive ‎‎graph ‎‎structure

Articles in Press

Document Type : Research Article

Authors

¹ Department of Applied Mathematics‎, ‎University of Kurdistan‎, ‎Sanandaj‎, ‎Iran

² Department of Applied Mathematics, University of Kurdistan, Sanandaj, Iran

³ School of Engineering‎, ‎RMIT University‎, ‎Melbourne‎, ‎Australia

10.22124/jmm.2025.30225.2714

Abstract

‎Feature selection is vital for improving high-dimensional data analysis by identifying a subset of representative ‎and‎ ‎uncorrelated features‎.
‎This paper presents ‎an unsupervised feature selection ‎algorithm‎ based on subspace learning ‎and ‎adaptive ‎g‎raph ‎structure ‎(UFSAG‎)‎‎.
‎The ‎UFSAG uses matrix factorization to preserve global data structure and incorporates local correlations into its objective function‎.
‎It also integrates sample similarity graph learning to maintain data geometry‎.
‎Unlike prior methods‎, ‎UFSAG employs adaptive local structure learning to reduce noise and enhance feature selection‎.
‎By inducing row sparsity in the feature coefficient matrix using the $\ell_{2,1}$-norm‎, ‎UFSAG identifies representative features‎.
‎Comparative experiments on six datasets show UFSAG's superior clustering performance over twelve state-of-the-art methods‎.

Keywords