Prediction where optimized. Previously for each data a difficulty was assigned (easy or hard) and was then given to the right classifier for prediction. Now, everything is sorted at once, then all easy data are predicted at once and the same for hard data. This results in a speed up between 10 and 20.
UnePatate5010
commented
4 months ago
Prediction where optimized. Previously for each data a difficulty was assigned (easy or hard) and was then given to the right classifier for prediction. Now, everything is sorted at once, then all easy data are predicted at once and the same for hard data. This results in a speed up between 10 and 20.