Patch-Based Low-Rank Matrix Completion for Learning of Shape and Motion Models from Few Training Samples.
Jan EhrhardtMatthias WilmsHeinz HandelsPublished in: ECCV (4) (2016)
Keyphrases
- low rank
- matrix completion
- training samples
- kernel matrix
- supervised learning
- learning algorithm
- convex optimization
- missing data
- rank minimization
- motion model
- linear combination
- singular value decomposition
- high dimensional data
- low rank matrix
- matrix factorization
- training set
- trace norm
- small number
- semi supervised
- feature space
- training data
- active learning
- expectation maximization
- nearest neighbor
- machine learning
- singular values
- em algorithm