Efficient and Robust Feature Selection via Joint ℓ2,1-Norms Minimization

TLDR

Feature selection is crucial in many machine learning and bioinformatics tasks, where efficient and robust methods are needed to extract meaningful features and remove noise. The authors propose a robust feature selection method that jointly minimizes l2,1‑norms in both the loss function and regularization. The approach uses a regression objective with l2,1‑norm loss and regularization, an efficient algorithm with proven convergence, and joint sparsity across data points, and is evaluated on six datasets. The method successfully identified genomic and proteomic biomarkers and demonstrated superior performance across six datasets.

Abstract

Feature selection is an important component of many machine learning applications. Especially in many bioinformatics tasks, efficient and robust feature selection methods are desired to extract meaningful features and eliminate noisy ones. In this paper, we propose a new robust feature selection method with emphasizing joint l2,1-norm minimization on both loss function and regularization. The l2,1-norm based loss function is robust to outliers in data points and the l2,1-norm regularization selects features across all data points with joint sparsity. An efficient algorithm is introduced with proved convergence. Our regression based objective makes the feature selection process more efficient. Our method has been applied into both genomic and proteomic biomarkers discovery. Extensive empirical studies are performed on six data sets to demonstrate the performance of our feature selection method.