A proposal of l1 regularized distance metric learning for high dimensional sparse vector space.
Kenta MikawaManabu KobayashiMasayuki GotoShigeichi HirasawaPublished in: SMC (2014)
Keyphrases
- vector space
- high dimensional
- distance metric learning
- low dimensional
- nonlinear dimensionality reduction
- similarity search
- metric learning
- distance metric
- manifold learning
- dimensionality reduction
- high dimensional data
- distance function
- euclidean space
- data points
- metric space
- image classification
- feature space
- gender classification
- distance measure
- dimension reduction
- locally linear embedding
- semidefinite programming
- semi supervised
- nearest neighbor
- input space
- least squares
- covariance matrices
- objective function
- riemannian manifolds
- parameter space
- multi task
- kernel learning
- feature vectors
- kernel function
- multi class
- sparse representation