On Tue, Dec 06, 2011 at 10:25:47PM +0100, Alexandre Gramfort wrote: > regarding the scaling by n_samples using estimators I am convinced the right > thing to do cf. my current PR to do this also on SVM models
I think that scaling by n_samples makes sense in the supervised learning context (we often do the equivalent thing where we take the mean, rather than the sum, over the unregularized training objective, making the regularization invariant to the size of the training set), however there is a disconnect between the dictionary learning notion of n_samples and the supervised estimator notion of n_samples, and the conflation of these two because one can be implemented by the other. To be precise, (and I hope I got this right lest I confuse things further), a sparse coding problem with K different training examples and L different input features and M sparse components corresponds to K independent lasso problems with L training examples each and M input features. In this case, scaling the penalty by the Lasso "n_samples" corresponds to scaling by the sparse coding "n_features", which I think you'll agree is a bit weird. David ------------------------------------------------------------------------------ Cloud Services Checklist: Pricing and Packaging Optimization This white paper is intended to serve as a reference, checklist and point of discussion for anyone considering optimizing the pricing and packaging model of a cloud services business. Read Now! http://www.accelacomm.com/jaw/sfnl/114/51491232/ _______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
