machine learning - How to determine the number of weak classifiers in ada boost -


i want ask question: how many weak classifiers normaly use in ada boost classification algorithm. example if have number of features 8000 features obtained feature detector haris, hog, sift or whatever, how determine number of weak classifiers nr of features. have thought of using 1 weak classifier each type of feature , make weighted sum in manner. afraid may overfit...

to parameters of model should cross-validation.

if size of database allows (you have sufficient samples) split learning set validation set. cross validation works way : learn ~75% of learning set , test score onto remaining ~25%. , different values of parameters , pick value lead higest score.

if size of database doesn't allow can k-fold cross-validation (i won't explain here can on wikipedia).

scikit-learn implements tool called gridsearch "automatically" if provide right things.

http://scikit-learn.org/stable/modules/grid_search.html


Comments

Popular posts from this blog

apache - Remove .php and add trailing slash in url using htaccess not loading css -

javascript - jQuery show full size image on click -