Retrain classifier with new distributed data to see if it improves accuracy