Events
FDS
Community Detection and the Hypergraph Stochastic Block Model
 |
Speaker: Ioana Dumitriu (UCSD) Professor University of California, San Diego Monday, November 11, 2024 4:00PM - 5:00PM 3:30 PM - Pre-talk meet and greet teatime at 219 Prospect Street, 13th floor. There will be light snacks and beverages in the kitchen area.
|
Abstract: Community detection in complex networks is one of the fundamental problems in unsupervised learning, with applications from recommender systems to gene classification. In order to design state-of-the-art algorithms for community detection, as well as to predict and offer guarantees for their real-world performance, one often employs stochastic network models. The Hypergraph Stochastic Block Model (HSBM) is a set of basic models that have a lot of built-in flexibility and whose parameters can be adapted to suit various real-life networks (from social to biological and neural). One crucial issue in the study of these models is represented by the “threshold” relationships among the defining parameters: these are sets of constraints (inequalities) which characterize the difficulty of the recovery problem and describe the best possible type of algorithmic performance on a network subject to these constraints. We will provide an overview of the problem, regimes of recovery, and survey literature, and then focus on partial, almost exact, and exact recovery for variants of the HSBM.
This represents joint work with Haixiao Wang and Yizhe Zhu.
Speaker bio: Ioana Dumitriu holds a BA in Mathematics from New York University (1999) and a PhD in Applied Mathematics from Massachusetts Institute of Technology (2003). Following a Miller Research Fellowship at University of California, Berkeley (2003-2006), she became a regular faculty member in the Department of Mathematics at University of Washington, Seattle (2006-2019). She joined UCSD in 2019.
Dumitriu’s research spans a number of areas that fall under the large umbrella of data science: from numerical linear algebra and scientific computing to stochastic eigenanalysis (a.k.a. random matrix theory), and from discrete probability and spectra of random graphs to applications in machine learning, specifically to clustering and community detection.
Website: https://math.ucsd.edu/people/profiles/ioana-dumitriu
Add To: Google Calendar | Outlook | iCal File
Submit an Event
Interested in creating your own event, or have an event to share? Please fill the form if you’d like to send us an event you’d like to have added to the calendar.