ECE 6350 - [Interpretable and Explainable Machine Learning]

Spring. Not offered: 2023-2024. Next offered: 2024-2025. 3 credits. Letter grades only.

Enrollment limited to: Cornell Tech students. Offered in New York City at Cornell Tech.

M. Sabuncu.

Machine learning (ML) is being increasingly used in high-stakes domains such as medicine and autonomous navigation. Furthermore, in many applications of machine learning, an important objective is to gain insights about the data and understand mechanisms, rather than simply compute predictions. In this graduate level course, we will aim to cover recent advances in the rapidly evolving field of interpretable and explainable ML. We will review seminal position papers of the field, define and discuss the concepts of model interpretability and explainability, study different classes of interpretable models (e.g., prototype-based approaches, sparse linear models, rule-based techniques, generalized additive models), post-hoc explanations (black-box explanations including counterfactual explanations and saliency maps), and explore the connections between interpretability and other related concepts such as causality.

Outcome 1: Students will learn to evaluate and design machine learning models in a way that goes beyond prediction performance

Outcome 2: Through the class project, students will obtain hands-on experience in an ML application where interpretability and explainability is important. They will get to demonstrate their work in a project presentation.

Outcome 3: Students will analyze the connections between closely-related concepts such as causality, actionability and interpretability.