Abstract:Linear discriminant analysis (LDA) represents a simple yet powerful technique for partitioning a p-dimensional feature vector into one of K classes based on a linear projection learned from N labeled observations. However, it is well-established that in the high-dimensional setting (p > N) the underlying projection estimator degenerates. Moreover, any linear discriminate function involving a large number of features may be difficult to interpret. To ameliorate these issues, two general categories of sparse LDA modifications have been proposed, both to reduce the number of active features and to stabilize the resulting projections. The first, based on optimal scoring, is more straightforward to implement and analyze but has been heavily criticized for its ambiguous connection with the original LDA formulation. In contrast, a second strategy applies sparse penalty functions directly to the original LDA objective but requires additional heuristic trade-off parameters, has unknown global and local minima properties, and requires a greedy sequential optimization procedure. In all cases the choice of sparse regularizer can be important, but no rigorous guidelines have been provided regarding which penalty might be preferable. Against this backdrop, we winnow down the broad space of candidate sparse LDA algorithms and promote a specific selection based on optimal scoring coupled with a particular, complementary sparse regularizer. This overall process ultimately progresses our understanding of sparse LDA in general, while leading to targeted modifications of existing algorithms that produce superior results in practice on three high-dimensional gene data sets.

Sr-LDA:Sparse and Reduced-Rank Linear Discriminant Analysis for High Dimensional Matrix

A New Reduced-Rank Linear Discriminant Analysis Method and Its Applications

Sparse linear discriminant analysis by thresholding for high dimensional data

Understanding and Evaluating Sparse Linear Discriminant Analysis

Computational and Theoretical Analysis of Supervised Dimensionality Reduction

Multimodal Linear Discriminant Analysis Via Structural Sparsity

L1 Least Squares for Sparse High-Dimensional LDA

Optimal Feature Selection for Sparse Linear Discriminant Analysis and Its Applications in Gene Expression Data

Linear Discriminant Regularized Regression

Sparse Local Discriminant Projections for Feature Extraction

Generalized robust linear discriminant analysis for jointly sparse learning

Robust Linear Discriminant Analysis Using Ratio Minimization of L1,2-Norms

SRDA: an Efficient Algorithm for Large-Scale

Adaptive Local Linear Discriminant Analysis

SRDA: an Efficient Algorithm for Large-Scale Discriminant Analysis

Sparse Trace Ratio LDA for Supervised Feature Selection

Regularized Deep Linear Discriminant Analysis

Efficient and Robust Sparse Linear Discriminant Analysis for Data Classification

Local Feature Discriminant Projection

Regularized Linear Discriminant Analysis Using a Nonlinear Covariance Matrix Estimator