Unified tooling for machine learning interpretability

Harsha Nori (Microsoft), Samuel Jenkins (Microsoft), Rich Caruana (Microsoft)

11:20am–12:00pm Wednesday, September 25, 2019

Location: 3B - Expo Hall

Data Science, Machine Learning, & AI, Expo Hall

Secondary topics: Ethics

Average rating:

(4.33, 3 ratings)

Who is this presentation for?

Data scientists, machine learning engineers, and software engineers

Level

Intermediate

Description

In machine learning, often a trade-off must be made between accuracy and intelligibility: the most accurate models usually are not very intelligible (e.g., random forests, boosted trees, and neural nets), and the most intelligible models usually are less accurate (e.g., linear or logistic regression). Interpretability research has focused on minimizing this trade-off by developing more accurate interpretable models and by developing new techniques to explain black box models.

Harsha Nori, Sameul Jenkins, and Rich Caruana walk you through a framework for thinking about interpretability and help you choose the right interpretability method for a variety of real-world tasks. They also present new tooling from Microsoft that helps with all forms of interpretability—both training accurate, interpretable models and understanding black box models. This toolkit includes the first Python implementation of a powerful new learning algorithm developed at Microsoft: GA2M.

Prerequisite knowledge

Experience training and evaluating machine learning systems

What you'll learn

Discover when and how to use a variety of machine learning interpretability methods through case studies of real-world situations
Learn how to use a new interpretability Python toolkit from Microsoft

Harsha Nori

Microsoft

Harsha Nori is a data scientist at Microsoft. He works on interpretability for machine learning.

Samuel Jenkins

Microsoft

Sameul Jenkins is a data scientist at Microsoft. He works on interpretability for machine learning.

Rich Caruana

Microsoft

Rich Caruana is a principal researcher at Microsoft Research. Previously, he was on the faculty in the Computer Science Department at Cornell University, at UCLA’s Medical School, and at Carnegie Mellon University’s Center for Learning and Discovery. Rich received an NSF CAREER Award in 2004 (for meta clustering); best paper awards in 2005 (with Alex Niculescu-Mizil), 2007 (with Daria Sorokina), and 2014 (with Todd Kulesza, Saleema Amershi, Danyel Fisher, and Denis Charles); co-chaired KDD in 2007 (with Xindong Wu); and serves as area chair for Neural Information Processing Systems (NIPS), International Conference on Machine Learning (ICML), and KDD. His research focus is on learning for medical decision making, transparent modeling, deep learning, and computational ecology. He holds a PhD from Carnegie Mellon University, where he worked with Tom Mitchell and Herb Simon. His thesis on multi-task learning helped create interest in a new subfield of machine learning called transfer learning.