Low-Rank Softmax Can Have Unargmaxable Classes in Theory but Rarely in Practice.
Andreas GrivasNikolay BogoychevAdam LopezPublished in: CoRR (2022)
Keyphrases
- low rank
- missing data
- linear combination
- matrix factorization
- low rank matrix
- convex optimization
- matrix completion
- rank minimization
- singular value decomposition
- kernel matrix
- semi supervised
- high dimensional data
- high order
- minimization problems
- matrix decomposition
- low rank matrices
- robust principal component analysis
- singular values
- data matrix
- affinity matrix
- collaborative filtering
- data analysis
- non rigid structure from motion
- pattern recognition
- feature selection