Rating is available when the video has been rented.
This feature is not available right now. Please try again later.
Published on Dec 2, 2015
Talk given by Joseph Salmon at CIMAT on November, 6th, during the Workshop on Image Processing/Statistical Pattern Recognition.
High dimensional regression benefits from sparsity promoting regularizations. Screening rules leverage the known sparsity of the solution by ignoring some variables in the optimization, hence speeding up solvers. When the procedure is proven not to discard features wrongly the rules are said to be "Safe". I will derive new safe rules for generalized linear models regularized with l1 and l1/l2 norms. GAP Safe rules can cope with any iterative solver and we illustrate their performance on coordinate descent for various applications (eg. multi-task Lasso, binary and multinomial logistic regression) demonstrating significant speed ups.
This is a joint work with E. Ndiaye, O. Fercoq and A. Gramfort (to appear in ICML 2015 and NIPS2015).