A Doubly Regularized Linear Discriminant Analysis Classifier With Automatic Parameter Selection

Linear discriminant analysis (LDA) based classifiers tend to falter in many practical settings where the training data size is smaller than, or comparable to, number of features. As a remedy, different regularized LDA (RLDA) methods have been proposed. These may still perform poorly depending on and quality available data. In particular, test deviation from model, for example, due noise contamination, can cause severe performance degradation. Moreover, these commit further Gaussian assumption (upon which established) tune their regularization parameters, compromise accuracy when dealing with real To address issues, we propose doubly classifier that denote as R2LDA. proposed R2LDA approach, RLDA score function converted into an inner product two vectors. By substituting expressions estimators vectors, obtain involves parameters. set values adopt three existing techniques; constrained perturbation approach (COPRA), bounded (BPR) algorithm, generalized cross-validation (GCV) method. are used parameters linear estimation models, sample covariance matrix's square root being operator. Results obtained both synthetic demonstrate consistency effectiveness especially scenarios involving contaminated not observed during phase.