Overview

In this presentation, I will first briefly introduce the main problems in trustworthy machine learning, namely interpretability, fairness, privacy, and adversarial machine learning. The goal of the first part of the presentation is to highlight the importance of Trustworthy Machine Learning and to introduce some materials for starting research on this topic. Then in the second half of the talk, I will explain in more detail one particular method for interpretation: symbolic metamodeling for interpreting black-boxes. In symbolic metamodeling, the goal is to find an interpretable approximation for a given black-box model (e.g. a neural net).