LIBSVM: MATLAB function "trainsvm" deprecated and replaced by "fitcsvm"

As MATLAB ha depreciated "svmtrain" and replaced it by "fitcsvm", LIBSVM is giving error:

% This is an example of using precomputed kernel
% using Libsvm in MATLAB, where K is the precomputed kernel (Gram matrix).

model = fitcsvm(TrainLabels, K, '-t 4');    

Error using internal.stats.parseArgs (line 42)
Wrong number of arguments.
Error in classreg.learning.paramoptim.parseOptimizationArgs (line 5)
[OptimizeHyperparameters,~,~,RemainingArgs] = internal.stats.parseArgs(...

Error in fitcsvm (line 312)
[IsOptimizing, RemainingArgs] =
classreg.learning.paramoptim.parseOptimizationArgs(varargin);

Error in myKernelLIBSVM (line 153)
        model = fitcsvm(TrainLabels, K, '-t 4');

Earlier, this would work fine using svmtrain.

Error using svmtrain (line 230)
svmtrain has been removed. Use fitcsvm instead.

Answer 1

[Walter Roberson]

You have

fitcsvm(TrainLabels, K, '-t 4')

The valid syntaxes for fitcsvm are

Mdl = fitcsvm(Tbl,ResponseVarName)
Mdl = fitcsvm(Tbl,formula)
Mdl = fitcsvm(Tbl,Y)
Mdl = fitcsvm(X,Y)
Mdl = fitcsvm(___,Name,Value)

The only one of this that permits a third parameter is the last one, with name/value pairs. But '-t 4' is not a name/value pair.

MATLAB's svmtrain() also never supported that syntax. That syntax was valid only for libsvm . You can still pull in libsvm if you want to.

The syntax for fitcsvm never has the label information as a separate first parameter, only integrated as part of the first parameter, or as a second parameter.

You should be having a look at the KernelFunction option