Application of double asymptotics and random matrix theory in error estimation of regularized linear discriminant analysis.
Amin ZollanvariEdward R. DoughertyPublished in: GlobalSIP (2013)
Keyphrases
- linear discriminant analysis
- error estimation
- discriminant analysis
- feature extraction
- dimensionality reduction
- dimension reduction
- face recognition
- feature space
- discriminant features
- principal component analysis
- discriminative information
- fisher criterion
- null space
- small sample size
- class separability
- computer vision
- high dimensional data
- kernel discriminant analysis
- support vector machine svm
- model selection
- neural network
- least squares
- pairwise
- pattern recognition
- similarity measure
- random matrix theory