Characterization of the equivalence of robustification and regularization in linear, median, and matrix regression.
Dimitris BertsimasMartin S. CopenhaverPublished in: CoRR (2014)
Keyphrases
- trace norm
- simple linear
- square loss
- regression model
- kernel ridge regression
- reproducing kernel hilbert space
- square root
- linear regression
- aggregating algorithm
- pseudo inverse
- linearly independent
- equivalence relationship
- linear systems
- regression function
- ridge regression
- regression algorithm
- model selection
- locally linear
- singular value decomposition
- piecewise linear
- eigenvalue decomposition
- canonical correlation analysis
- regularization parameter
- rank minimization
- gradient boosting
- kernel matrices
- support vector
- cross validation
- closed form
- low rank
- matrix valued
- nonlinear regression
- regression problems
- partial least squares
- regression method
- semidefinite