Interpretability in Machine Learning: An Overview

This essay provides a broad overview of the sub-field of machine learning interpretability. While not exhaustive, my goal is to review conceptual frameworks, existing research, and future directions. I follow the categorizations used in Lipton et al.'s Mythos of Model Interpretability, which I think is the best paper for understanding the different definitions of interpretability. We'll go over many ways to formalize what "interpretability" means. Broadly, interpretability focuses on the how. It's focused on getting some notion of an explanation for the decisions made by our models. Below, each section is operationalized by a concrete question we can ask of our machine learning model using a specific definition of interpretability. If you're new to all this, we'll first briefly explain why we might care about interpretability at all.