According to paper By Sun, Ji an Ye; A Least Squares Formulation for Canonical Correlation Analysis http://www.machinelearning.org/archive/icml2008/papers/270.pdf CCA can be reformulated as a least squares problem, where the L1 penalty can be added, and then it can be solved by LARS (least angle regression) which has a benefit for computing the whole solution path for all levels of regularization for a cost of fitting only one model.

I am bit lost in the math, so how is this possible? Is this true for any solver that can solve linear regression with l1 regularization? And most important, how can I reformulate my CCA problem, so it can be solved with LARS (and therefore for all regularization parameters at once)?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.