Sparse Coding Models Can Exhibit Decreasing Sparseness while Learning Sparse Codes for Natural Images.
Joel ZylberbergMichael Robert DeWeesePublished in: PLoS Comput. Biol. (2013)
Keyphrases
- sparse coding
- natural images
- dictionary learning
- sparse codes
- deep belief networks
- unsupervised learning
- discriminative dictionary
- image patches
- discriminative power
- image classification
- sparse representation
- multiscale
- higher order
- input image
- denoising
- reconstruction error
- image statistics
- linear combination
- object recognition
- prior knowledge
- learning algorithm
- image representation
- natural image statistics
- supervised learning
- active learning
- feature space
- data analysis
- test images
- bag of words
- reinforcement learning
- pairwise
- high resolution