Regularizers versus Losses for Nonlinear Dimensionality Reduction: A Factored View with New Convex Relaxations.
James NeufeldYaoliang YuXinhua ZhangRyan KirosDale SchuurmansPublished in: ICML (2012)
Keyphrases
- nonlinear dimensionality reduction
- convex relaxation
- manifold learning
- dimensionality reduction
- high dimensional data
- low dimensional
- convex optimization
- globally optimal
- data sets
- riemannian manifolds
- total variation
- linear combination
- state space
- multi label
- high dimensional
- locally linear embedding
- optimization methods
- dimensionality reduction methods
- multistage
- multiple kernel learning
- text categorization
- signal processing
- face recognition
- image segmentation
- neural network