OptShrink: An Algorithm for Improved Low-Rank Signal Matrix Denoising by Optimal, Data-Driven Singular Value Shrinkage.
Raj Rao NadakuditiPublished in: IEEE Trans. Inf. Theory (2014)
Keyphrases
- singular values
- low rank
- matrix decomposition
- singular value decomposition
- denoising
- matrix completion
- learning algorithm
- trace norm
- convex optimization
- closed form
- fundamental matrix
- semi supervised
- similarity measure
- missing data
- matrix factorization
- high dimensional data
- kernel matrix
- low rank approximation
- feature vectors
- image segmentation