How to declare kfun function in svmtrain

2 views (last 30 days)
Ikra89
Ikra89 on 13 Jun 2013
Hi all
i want to train data using polynomial kernel which have:
gamma, C, r, d.
svmtrain only has boxConstraint (C) and Polyorder (d)
so i have to declare function by my self using @kfun
i've tried many times but still got "*Function definitions are not permitted in this context.*"
where i must put this code?
function K = kfun(U,V,gamma,r)
K = gamma*(U*V')+r;
in command line and in .m script file I still got error like that
anybody can help me?
thanks before
  1 Comment
Ahmed
Ahmed on 13 Jun 2013
You are probably trying to define a function in a script file. This is not possible. Put your function in a separate m-File called kfun.m or define an anonymous function.
kfun = @(U,V,gamma,r) gamma*(U*V')+r;

Sign in to comment.

Answers (1)

Ahmed
Ahmed on 6 Jul 2013
Edited: Ahmed on 6 Jul 2013
data = randn(100,10);
groups = rand(100,1)>0.5;
g = 1;
r = 1;
svmstruct = svmtrain(data, groups, ...
'Kernel_Function',@(U,V) g*(U*V')+r,'BoxConstraint', 0.2);
(note that 'gamma' is pre-defined function, therefore you should better use another variable name)

Products

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!