Neural Rank Collapse: Weight Decay and Small Within-Class Variability Yield Low-Rank Bias.
Emanuele ZangrandoPiero DeiddaSimone BrugiapagliaNicola GuglielmiFrancesco TudiscoPublished in: CoRR (2024)
Keyphrases
- low rank
- frobenius norm
- linear combination
- missing data
- convex optimization
- nuclear norm
- singular values
- matrix factorization
- low rank matrix
- singular value decomposition
- matrix completion
- rank minimization
- semi supervised
- low rank approximation
- high order
- high dimensional data
- kernel matrix
- matrix decomposition
- small number
- minimization problems
- neural network
- trace norm
- non rigid structure from motion
- data matrix
- data sets
- low rank matrices
- low rank and sparse