prtKernel Base class for prtKernel objects.
prtKernel is the base class for all prtKernel objects. It is an
abstract class and should not be instantiated. All prtKernel
objects implement the following methods:
kernel = kernel.train(dataSet) Train the kernel using the specified
dataSet. This builds a trained kernel object that can be run.
yOut = kernel.run(dataSet) Run a trained kernel object on the data
in dataSet and output the resulting gram matrix in
kernel = and(kernel1,kernel2) Combine two kernels into a
prtKernelSet (also a prtKernel). This is used to join multiple
kernels together. Unlike the plus operation for combining
prtActions, the AND operation trains and runs each kernel
individually on the provided data.
Note, cross validation is not a valid operation on prtKernel
objects, and therefore the CROSSVALIDATE and KFOLDS operations are
% Example syntax of the AND method:
kernels = prtKernelDc & prtKernelRbf; % kernels is prtKernelSet