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.

Varying Coefficient Linear Discriminant Analysis for Dynamic Data

Dynamic Linear Discriminant Analysis in High Dimensional Space

Linear Discriminant Regularized Regression

Adaptive and Fuzzy Locality Discriminant Analysis for Dimensionality Reduction

Feature extraction by dynamically revising the discriminant information in the objective space

Polynomial Linear Discriminant Analysis

MBLDA: A Novel Multiple Between-Class Linear Discriminant Analysis

Penalized Discriminant-Analysis

Linear discriminant analysis based on genetic algorithm

Locality Adaptive Discriminant Analysis Framework

Multimodal Linear Discriminant Analysis Via Structural Sparsity

Multiclass Discriminant Analysis Via Adaptive Weighted Scheme

The Dantzig Discriminant Analysis with High Dimensional Data

Random-matrix Regularized Discriminant Analysis of High-Dimensional Dataset

Least Square Incremental Linear Discriminant Analysis

Understanding and Evaluating Sparse Linear Discriminant Analysis

Adaptive Local Linear Discriminant Analysis

Sparse linear discriminant analysis by thresholding for high dimensional data

Generalized Discriminant Analysis: A Matrix Exponential Approach

Training Linear Discriminant Analysis in Linear Time

A procedure of linear discrimination analysis with detected sparsity structure for high-dimensional multi-class classification