In this paper, we propose an improved linear discriminant analysis, called spectrally-corrected and regularized linear discriminant analysis (SCRLDA). This method integrates the design ideas of the sample spectrally-corrected covariance matrix and the regularized discriminant analysis. The SCRLDA method is specially designed for classification problems under the assumption that the covariance matrix follows a spiked model. Through the real and simulated data analysis, it is shown that our proposed classifier outperforms the classical R-LDA and can be as competitive as the KNN, SVM classifiers while requiring lower computational complexity.

Spectrally-Corrected and Regularized Linear Discriminant Analysis for Spiked Covariance Model