We indeed need to find a way to explicitly handle the precomputed "kernel", "Gram", "affinity" matrix in a consistent way.
Maybe it would be better to have a dedicated method for this use case rather that using complex kwargs in the `fit` method. What about using a `fit_from_kernel` or `fit_from_affinity` method (if you find a better name)? If someone wants to make a design proposal that is able to handle the following use cases with worked examples: - precomputed kernels for SVM both for classification and regression - precomputed affinity for spectral clustering and affinity propagation - precomputed empirical covariance for GraphLasso - (maybe there are other similar cases in Manifold) - how this would impact pipelining, cross validation and grid search. -- Olivier ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________ Scikit-learn-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
