Training invariances and the low-rank phenomenon: beyond linear networks.
Thien LeStefanie JegelkaPublished in: ICLR (2022)
Keyphrases
- low rank
- low rank matrix
- missing data
- convex optimization
- linear combination
- matrix factorization
- singular value decomposition
- matrix completion
- kernel matrix
- rank minimization
- semi supervised
- high dimensional data
- matrix decomposition
- high order
- trace norm
- low rank matrices
- robust principal component analysis
- collaborative filtering
- supervised learning
- approximation methods
- network structure
- biological networks
- learning tasks
- singular values
- linear svm
- training samples
- pattern recognition
- neural network