Fisher core relational implementation

Question

I read an article on the Fisher Relational Core , which includes Bayesian logic to calculate the Fisher score, and then uses SVM to get class labels for each data item.

I do not have a strong background from machine learning. Can someone please let me know how to start implementing the end-to-end Relational Fisher core and what contribution he will expect? I could not find an easy step-by-step flow showing this implementation. I agree with the use of libraries for SVM, etc. (For example, libsvm), but I would like to know the end-to-end stream (as light as possible language). Any help would be greatly appreciated.

Answer 1

libsvm does not implement the Fisher Communication Core, however you can calculate the Fisher information matrix as described in the document and use it as a pre-computed kernel input in libsvm . See using pre-computed kernels with libsvm