[Colloquium] David Gleich: " Topological Structure of Complex Prediction and Graphs"

[Roberto Vale]

UNIVERSITY OF CHICAGO
COMPUTER SCIENCE DEPARTMENT
PRESENTS

David Gleich
Professor and University Faculty Scholar
Department of Computer Science
Purdue University


[david gleich.jpeg]


Tuesday, May 7th
2:30pm - 3:30pm
In Person: John Crerar Library 390

Title: Topological Structure of Complex Prediction and Graphs

Abstract: It is now standard practice across science to study models that have been trained, fit, or learned based on a set of data. Many of these models involve a large number of parameters that make direct interpretation of the model challenging and a near black-box model view appropriate. We explore the possibilities of using ideas based on topological analysis methods to understand and evaluate these complex prediction functions. These show a surprising ability to generate easy to understand insights into these black boxes. In addition to this deep dive on analyzing prediction methods, I’ll discuss recent work on validating a decade-old structural prediction in terms of the conductance of sets in graphs by Leskovec, Lang, Dasgupta, and Mahoney. This is based on https://arxiv.org/abs/2207.14358<https://urldefense.com/v3/__https:/arxiv.org/abs/2207.14358__;!!BpyFHLRN4TMTrA!4GP-I3pEEuwamINV8kVIwZzQjqeAytiDNQ0uvcLzeWli8KN7vasZR25Rdl7SuGLVrBq6Tgv-p_clPYOYOzHoGjU$> and https://arxiv.org/abs/2303.14550<https://urldefense.com/v3/__https:/arxiv.org/abs/2303.14550__;!!BpyFHLRN4TMTrA!4GP-I3pEEuwamINV8kVIwZzQjqeAytiDNQ0uvcLzeWli8KN7vasZR25Rdl7SuGLVrBq6Tgv-p_clPYOYRKYAPs0$>. For software, see https://github.com/MengLiuPurdue/Graph-Topological-Data-Analysis<https://urldefense.com/v3/__https:/github.com/MengLiuPurdue/Graph-Topological-Data-Analysis__;!!BpyFHLRN4TMTrA!4GP-I3pEEuwamINV8kVIwZzQjqeAytiDNQ0uvcLzeWli8KN7vasZR25Rdl7SuGLVrBq6Tgv-p_clPYOYfqgoNKE$>, https://mengliupurdue.github.io/Graph-Topological-Data-Analysis/<https://urldefense.com/v3/__https:/mengliupurdue.github.io/Graph-Topological-Data-Analysis/__;!!BpyFHLRN4TMTrA!4GP-I3pEEuwamINV8kVIwZzQjqeAytiDNQ0uvcLzeWli8KN7vasZR25Rdl7SuGLVrBq6Tgv-p_clPYOY0uEgeZA$>, https://github.com/luotuoqingshan/mu-conductance-low-rank-sdp<https://urldefense.com/v3/__https:/github.com/luotuoqingshan/mu-conductance-low-rank-sdp__;!!BpyFHLRN4TMTrA!4GP-I3pEEuwamINV8kVIwZzQjqeAytiDNQ0uvcLzeWli8KN7vasZR25Rdl7SuGLVrBq6Tgv-p_clPYOYnX0o-ho$>

Bio: David Gleich is a Professor and University Faculty Scholar in the Computer Science Department at Purdue University whose research is on novel models and fast large-scale algorithms for data-driven computing including network analysis and applications in bioinformatics. He is committed to making software available based on this research and has written software packages such as MatlabBGL with thousands of users worldwide. Gleich has received a number of awards for his research including a SIAM Outstanding Publication prize (2018), a Sloan Research Fellowship (2016), an NSF CAREER Award (2011), the John von Neumann post-doctoral fellowship at Sandia National Laboratories in Livermore CA (2009). His research has been funded by the NSF, DOE, DARPA, IARPA, and NASA.
For more information, see his website: https://www.cs.purdue.edu/homes/dgleich/<https://urldefense.com/v3/__https:/www.cs.purdue.edu/homes/dgleich/__;!!BpyFHLRN4TMTrA!4GP-I3pEEuwamINV8kVIwZzQjqeAytiDNQ0uvcLzeWli8KN7vasZR25Rdl7SuGLVrBq6Tgv-p_clPYOYUOMoDm4$>

Host: Michael Maire

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20240429/46eb3798/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: david gleich.jpeg
Type: image/jpeg
Size: 3021 bytes
Desc: david gleich.jpeg
URL: <http://mailman.cs.uchicago.edu/pipermail/colloquium/attachments/20240429/46eb3798/attachment-0001.jpeg>