VLFeat SVM storage

Question

I am working with VLFeat and I have code that trains a linear SVM on my dataset. I would like to save the SVM to a file somehow so that I can load it up later and test it on several other datasets. What is the best way to do this?

edit: I am using the C++ API. Is it enough to save the model by using vl_svm_get_model? Would a simple serialization of the bytes of the svm struct work?

score -1 · Accepted Answer · answered May 02 '14 at 19:59

-1

I am not using this approach however it should be sufficient to save the model and bias terms to a file for later testing

answered May 02 '14 at 19:59

Max Ehrlich

2,479
1
32
44

2

Dont forget the bias multiplier! – remi May 12 '14 at 09:21

VLFeat SVM storage

1 Answers1