Feedback on Improving Human Behavior Prediction in Simultaneous-Move Games