Discriminative Subspace Learning

Discriminative subspace learning seeks a projection that best separates the data in a least-squares sense. It requires labeled data for training. It is a supervised technique.

Conventional Discriminative Subspace Learning

1. Linear Discriminant Analysis (LDA) (Fisher, 1936) [Euclidean structure]
2. Canonical Correlation Analysis (CCA) (Hotelling, 1936)
3. Maximum Margin Criterion (MMC) (Li, 2004)

Recent Discriminative Subspace Learning

1. Quadratic Discriminant Analysis (QDA) ()
2. Regularized Discriminant Analysis (RDA) (Guo et al., 2005)