A=ADABOOST(C,H) returns a template object initialized with learning algorithm C and hyperparameters H.
The template object is a simple linear support vector machine, with
soft margin hyperparameter C. Please use this template to
implement your own algorithms.
Hyperparameters, and their defaults
kmax = 5 -- number of weak learners
Methods:
train, test
Model:
child -- learning algorithm
Example:
Use adaboost with 1-knn as weak learner and validate with 2 fold cross validation.
c1=[2,0];
c2=[-2,0];
X1=randn(50,2)+repmat(c1,50,1);
X2=randn(50,2)+repmat(c2,50,1);
d=data([X1;X2],[ones(50,1);-ones(50,1)]);
[r,a]=train(cv(adaboost(knn),'folds=2'),d);
|
Reference : The boosting approach to machine learning: An overview |
|
Author : Robert E. Schapire |