function [final_class] = k_adaboost_evaluate(patterns, classes, B_t)

%Use majority voting to classify each point (use tree weights)
T = length(classes);
for t = 1:T
[labels(:,t)] = k_classify_w_tree( patterns, classes{ t } ).';
% 1-mean(labels(:,t)==targets)
end

weight = log(1./B_t);

for r = 1:size(labels,1)
    ind_0 = find(labels(r,:) == 0);
    ind_1 = find(labels(r,:) == 1);
    weight_0 = (sum(weight(ind_0)));
    weight_1 = (sum(weight(ind_1)));
    if weight_0 > weight_1
        final_class(r,1) = 0;
    else
        final_class(r,1) = 1;
    end
end