Once every week while school is in session, EECS invites a distinguished researcher or practitioner in a computer science or electrical and computer engineering-related field to present their ideas and/or work. Talks are generally targeted to electrical engineering and computer science graduate students. This colloquium series is free and open to everyone.

Upcoming Colloquia

Robust Methods for Topology Estimation in Unsupervised Learning

Monday, February 24, 2020 - 4:00pm to 4:50pm
Linus Pauling Science Center 125

Speaker Information

Shay Deutsch
Assistant Adjunct Professor
Mathematics Department
University of California, Los Angeles

Abstract

Learning graph connectivity has broad-ranging applications from 3D reconstruction to unsupervised learning. In this talk I will introduce a new method to learn the graph structure underlying noisy point set observations assumed to lie near a complex manifold. Rather than assuming regularity of the manifold itself, as customary, we assume regularity of the geodesic flow through the boundary of arbitrary regions on the graph. The idea is to exploit this more flexible notion of regularity, captured by the discrete equivalent of the isoperimetric inequality for closed manifolds, to infer the graph structure.

In a broader perspective, when studying the topology of the graph networks, we would like to learn new representations that capture not only local connectivity, i.e., nodes that belong to the same local structure, but also similarity which is based on their structural role in the graph. I will discuss a new approach and vision towards learning a good trade-off between these local and structural types of similarities that includes diverse possible applications including point clouds, biological networks and social networks.

Speaker Bio

Shay Deutsch received a B.Sc. in Mathematics from the Technion—Israel Institute of Technology in 2007, an M.Sc. in Applied Mathematics from Tel Aviv University in 2010, and a Ph.D. in Computer Science from the University of Southern California (USC) in 2016. He is currently an Assistant Adjunct Professor in the Mathematics Department at the University of California, Los Angeles (UCLA). His research work is in the union of transfer learning, graph signal processing and graph networks, where his research is dedicated to developing robust methods for unsupervised learning. His most recent research efforts focus on developing cohesive relations between embedding topology and graph networks using uncertainty principles on graphs.

Friends Don’t Let Friends Deploy Black-Box Models: The Importance of Intelligibility in Machine Learning

Monday, March 2, 2020 - 4:00pm to 4:50pm
Linus Pauling Science Center 125

Speaker Information

Rich Caruana
Senior Principal Researcher
Microsoft

Abstract

In machine learning often a tradeoff must be made between accuracy and intelligibility: the most accurate models usually are not very intelligible, and the most intelligible models usually are less accurate.  This often limits the accuracy of models that can safely be deployed in mission-critical applications such as healthcare where being able to understand, validate, edit, and ultimately trust a model is important.  We have developed a learning method based on generalized additive models (GAMs) that is as accurate as full complexity models, but even more intelligible than linear models.  This makes it easy to understand what a model has learned and to edit the model when it learns inappropriate things.  In the talk I’ll present several case studies where these high-accuracy GAMs discover surprising patterns in the data that would have made deploying a black-box model risky. I’ll also show how we’re using these models to uncover bias in models where fairness and transparency are important.  Every data set is flawed in surprising ways --- you need intelligibility.

 

Speaker Bio

Rich Caruana is a Senior Principal Researcher at Microsoft.  His research focus is on intelligible/transparent modeling, machine learning for medical decision making, deep learning, and computational ecology.  Before joining Microsoft, Rich was on the faculty in Computer Science at Cornell, at UCLA's Medical School, and at CMU's Center for Learning and Discovery.  Rich's Ph.D. is from CMU, where he worked with Tom Mitchell and Herb Simon.  His thesis on Multitask Learning helped create interest in a subfield of machine learning called Transfer Learning.  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), and co-chaired KDD in 2007 with Xindong Wu. 

 

Past Colloquia

Robert Holte
Monday, June 3, 2019 - 4:00pm to 4:50pm
Giuseppe Raffa
Monday, May 20, 2019 - 4:00pm to 4:50pm
Kannan A Sankaragomathi
Monday, May 13, 2019 - 4:00pm to 4:50pm
Monday, May 13, 2019 - 10:00am to 11:00am
Subhasish Mitra
Monday, May 6, 2019 - 4:00pm to 4:50pm
Kejun Huang
Monday, April 15, 2019 - 4:00pm to 4:50pm
Bo Li
Friday, April 12, 2019 - 10:00am to 11:00am

Pages