Seminar Series - Network Theory
UC Davis is hosting an exciting series of talks and visits in association with the Network Theory HIP program. The speakers are accomplished scholars in network theory, with diverse research spanning computer science, physics, and data science. Refreshments and time for conversation will immediately follow after the talks. Please join us so that we can highlight to these accomplished individuals the vigor of interdisciplinary science activities ongoing across the UC Davis campus. Please email Dr. Raissa D'Souza if you would like to arrange an individual research meeting with any of the speakers.
Thursday, March 15
Speaker: Soheil Feizi, Postdoc at Stanford University Title: "Unsupervised Learning Revisited" 12:10-1pm in 1131 Kemper
Please contact Prof. D'Souza if you are interested in meeting the speaker.
Abstract: Modern datasets are massive, complex and often unlabeled. These attributes make unsupervised learning important in several data-driven application domains. This ever-growing area, although demonstrating excellent empirical performance, suffers from a major drawback. Unlike for classical learning methods, there is a lack of fundamental understanding of several modern approaches, hindering development of principled methods for improvement.
To resolve this issue, one approach is to draw appropriate connections between modern and classical learning methods. Thus, leveraging from the vast body of classical results, one can either develop new learning algorithms or improve existing ones in a principled way. In this talk, I am going to illustrate the success of this approach in two unsupervised learning problems, namely (1) learning a nonlinear dimensionality reduction of the data, and (2) learning probabilistic models from the data. In the first problem, by drawing connections with Maximal Correlation and PCA, our approach produces a new method called Maximally Correlated PCA, a nonlinear generalization of PCA with a data-dependent nonlinearity. In the second problem, by drawing connections to optimal transport, supervised learning and rate-distortion theory, our approach leads to a principled design of Generative Adversarial Networks (GANs) in a baseline scenario.
Bio: Soheil Feizi is a post-doctoral research scholar at Stanford University. He received his Ph.D. in Electrical Engineering and Computer Science (EECS) with a minor degree in Mathematics from the Massachusetts Institute of Technology (MIT). His research interests are in the area of machine learning, statistical inference and network science. He completed a M.Sc. in EECS at MIT, where he received the Ernst Guillemin award for his thesis, as well as the Jacobs Presidential Fellowship and the EECS Great Educators Fellowship. He also received the best student award at Sharif University of Technology from where he obtained his B.Sc.
Network Theory Seminar Series:
Tues, Feb 20, 3:10pm 1131 Kemper Hall Maksim Kitzak Department of Physics & Network Science Institute, Northeastern University Title: "Latent geometry in networked systems: theory and applications"
Thurs, Mar 1, 3:10pm 1131 Kemper Hall Peter Dodds Director of Vermont Complex Systems Center, University of Vermont Title: "Building Lexical Meters to explore Happiness, Health, Language, Public Opinion, and Stories"
Tues, Mar 6, 3:10pm 1131 Kemper Hall Thilo Gross Faculty of Engineering, University of Bristol Title: "Pattern formation in meta-food-webs"
Tues, Mar 13, 3:10pm 1131 Kemper Hall Marta Gonzalez Department of City and Regional Planning, UC Berkeley
Thurs, Mar 15, 12:10pm 1131 Kemper Hall Soheil Feizi Electrical Engineering, Stanford University