Dual Regularized Unsupervised Feature Selection Based on Matrix Factorization and Minimum Redundancy with application in gene selection

Gene expression data have become increasingly important in machine learning and computational biology over the past few years. In field of gene analysis, several matrix factorization-based dimensionality reduction methods been developed. However, such can still be improved terms efficiency reliability. this paper, an innovative approach to feature selection, called Dual Regularized Unsupervised Feature Selection Based on Matrix Factorization Minimum Redundancy (DR-FS-MFMR), is introduced. The major focus DR-FS-MFMR discard redundant features from set original features. order reach target, primary selection problem defined two aspects: (1) factorization weight representation matrix, (2) correlation information related selected set. Then, objective function enriched by employing characteristics along with inner product regularization criterion perform both redundancy minimization process sparsity task more precisely. To demonstrate proficiency method, a large number experimental studies are conducted nine datasets. obtained results indicate productivity for task.