Human-in-the-Loop Machine Learning

appendix  Machine learning refresher

This appendix covers the basics of machine learning that are most relevant to human-in-the-loop machine learning, including interpreting the output from a machine learning model; understanding softmax and its limitations; calculating accuracy through recall, precision, F-score area under the ROC curve (AUC), and chance-adjusted accuracy; and measuring the performance of machine learning from a human perspective. This book assumes that you have basic machine learning knowledge. Even if you are experienced, you may want to review this appendix. In particular, the parts related to softmax and accuracy are especially important for this book and are sometimes overlooked by people who are looking only at algorithms. ...