Model Selection in Kernel Methods Based on a Spectral Analysis of Label Information.
Mikio L. BraunTilman LangeJoachim M. BuhmannPublished in: DAGM-Symposium (2006)
Keyphrases
- spectral analysis
- model selection
- kernel methods
- label information
- semi supervised
- dimensionality reduction
- machine learning
- cross validation
- training data
- feature space
- distance metric
- class labels
- labeled data
- kernel function
- unlabeled data
- learning problems
- parameter estimation
- transfer learning
- support vector
- feature selection
- support vector machine
- kernel matrix
- hyperparameters
- mixture model
- high dimensional data
- learning tasks
- active learning
- geometric structure
- gaussian process
- unsupervised learning
- high dimensional
- autoregressive
- target domain
- generalization bounds
- linear discriminant analysis
- principal component analysis
- pairwise
- pattern recognition
- test data
- least squares
- data points
- learning algorithm
- data mining
- kernel machines
- data sets