Discriminative regularization of the latent manifold of variational auto-encoders.
Ingo KossykZoltan-Csaba MartonPublished in: J. Vis. Commun. Image Represent. (2019)
Keyphrases
- latent space
- category labels
- variational formulation
- manifold structure
- manifold regularization
- semi supervised
- low dimensional
- variational framework
- gaussian process latent variable models
- projection operator
- graph laplacian
- lower dimensional
- latent variables
- manifold learning
- robust statistics
- regularization parameter
- image segmentation
- parameter space
- total variation regularization
- euclidean space
- high dimensional
- log linear models
- optical flow
- matrix factorization
- gaussian process
- feature extraction
- feature space
- latent variable models
- regularization method
- dimensionality reduction
- generative model
- unsupervised learning
- semi supervised learning
- variational methods
- transfer learning
- smoothness term
- information geometry
- feature selection
- level set
- regularization methods
- free energy
- video compression
- energy functional
- high dimensional data
- kernel machines
- fisher information
- data points
- discriminative learning
- discriminative models
- probabilistic latent semantic analysis
- image restoration