On the eigenspectrum of the gram matrix and the generalization error of kernel-PCA.
John Shawe-TaylorChristopher K. I. WilliamsNello CristianiniJaz S. KandolaPublished in: IEEE Trans. Inf. Theory (2005)
Keyphrases
- generalization error
- gram matrix
- kernel pca
- cross validation
- principal component analysis
- upper bound
- model selection
- learning algorithm
- active learning
- kernel methods
- training data
- training set
- kernel principal component analysis
- input space
- dimensionality reduction
- feature space
- supervised learning
- kernel function
- sample size
- kernel matrix
- face recognition
- spectral clustering
- classification accuracy
- feature extraction
- component analysis
- machine learning
- training examples
- semi supervised learning
- data sets
- input data
- support vector machine
- positive definite