A feasible k-means kernel trick under non-Euclidean feature space.
Robert A. KlopotekMieczyslaw A. KlopotekSlawomir T. WierzchonPublished in: Int. J. Appl. Math. Comput. Sci. (2020)
Keyphrases
- kernel trick
- feature space
- k means
- kernel methods
- maximum margin
- input space
- dimensionality reduction
- kernel pca
- high dimensional feature space
- kernel function
- infinite dimensional
- spectral clustering
- regression algorithm
- feature vectors
- sparse coding
- kernel based nonlinear
- dimensionality reduction methods
- nonlinear models
- clustering algorithm
- euclidean distance
- clustering method
- feature extraction
- feature selection
- high dimensional
- training samples
- hyperplane
- input data
- structured output
- kernel learning
- classification accuracy
- machine learning
- data clustering
- principal component analysis
- euclidean space
- riemannian manifolds
- expectation maximization
- linear transformation
- training set
- data points
- image retrieval
- feature set
- low dimensional
- document clustering
- support vector machine
- multiple kernel learning
- discriminative learning
- metric learning
- training data
- sparse representation