"Srey " <firstname.lastname@example.org> wrote in message news:email@example.com... > Error using ==> svmtrain at 453 > Error calculating the kernel function: > Out of memory. Type HELP MEMORY for your options. > > Error in ==> create_train_test_data svmStruct = svmtrain(data,class); >
First, please type
to verify that you are using svmtrain from the official MATLAB installation.
You don't say what release you are using. In older MATLAB releases, the default solver for svmtrain was QP. This solver creates a full kernel matrix of size N-by-N for N observations. Since you have so many observations, you would run out of memory at this step.
Use the SMO solver instead. This is the default choice in recent MATLAB versions. The 'kernelcachelimit' parameter for SMO lets you control the size of the cached kernel matrix. If you set this parameter to a small value, svmtrain may take a long time to finish. If you set it to a large value, you may run out of memory. -Ilya