RobustBoost Binary Learner - Which Binary Learner to pick when using RobustBoost for Multiclass Problems (predict(fitececoc(...

2 次查看（过去 30 天）

Tim 2015-7-25

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/231143-robustboost-binary-learner-which-binary-learner-to-pick-when-using-robustboost-for-multiclass-prob

Hello everybody, since I try to use the "predict" function with my built fitcecoc model

tTree = templateTree('surrogate','on');
t = templateEnsemble('RobustBoost',300,tTree,'RobustErrorGoal',0.15);
result=sum(cTest~=predict(fitcecoc(fTrain,cTrain,'Coding','onevsall','Learners',t,'Type', 'Classification','Options',options),fTest));

The error code tells me to pick a BinaryLoss Function. On this page: http://de.mathworks.com/help/stats/compactclassificationecoc.predict.html however there is no suggestion for the ROBUSTBOOST algorithm. Does anybody know which one is suitable? (I think, but I am not sure at all!, that it should be a non-convex one)

Thanks everybody!