Improving Machine Learning with Human Feedback