Two-layer contractive encodings for learning stable nonlinear features.
Hannes SchulzKyungHyun ChoTapani RaikoSven BehnkePublished in: Neural Networks (2015)
Keyphrases
- feature extraction
- learning algorithm
- learning process
- learning systems
- prior knowledge
- learning problems
- feature space
- low level
- supervised learning
- online learning
- training examples
- orders of magnitude
- learning tasks
- extracted features
- data sets
- multi layer
- background knowledge
- co occurrence
- high dimensional
- support vector