Convex Calibrated Surrogates for Low-Rank Loss Matrices with Applications to Subset Ranking Losses.
Harish G. RamaswamyShivani AgarwalAmbuj TewariPublished in: NIPS (2013)
Keyphrases
- low rank
- convex optimization
- low rank matrix
- nuclear norm
- matrix completion
- singular value decomposition
- hinge loss
- data matrix
- matrix decomposition
- singular values
- positive semidefinite
- frobenius norm
- low rank and sparse
- low rank matrices
- matrix factorization
- eigendecomposition
- kernel matrix
- minimization problems
- missing data
- linear combination
- convex relaxation
- low rank approximation
- affinity matrix
- rank minimization
- interior point methods
- semi supervised
- high order
- rank aggregation
- pairwise comparison
- primal dual
- least squares
- trace norm
- norm minimization
- risk minimization
- high dimensional data
- semidefinite programming
- total variation
- feature subset
- non rigid structure from motion
- loss function
- binary matrices
- principal component analysis