Algorithm 1 Pseudo-code for error regularization and data optimization

[YPred, score] = classify(DeepNNx, ValidationX) [Index, val] = find(YPred ~= ValidationY) for n = [index]    if max(score(n) < 0.5)      ValidationX(n) = [];      ValidationY(n) = [];       elseif max(score(n) > 0.5)      ValXSwap(n) = ValidationX(n);       for val(n)~=ValidationY(n)          find(TrainingX(n)==ValidationY(n))       end     TraXSwap(n) = TrainingX(n);      ValidationX(n) = TraXSwap(n);      TrainingX(n) = ValXSwap(n);     else       …, fitcknn(ValidationX(n), ValidationY(n), …), …     end end