On Detecting Communities in Networks Using Spectral Graph Methods
Date: Thu, March 23, 2017
Time: 4:00pm
Location: Holmes 388
Speaker: Dr. Ambedkar Dukkipati
Abstract:
Spectral graph methods for detecting communities in networks have been very successful for two reasons: (i) good practical solutions and easy implementations, (ii) theoretical guarantees. In this talk, I will introduce these methods by giving details of underlying spectral graph theory results. I will also talk about theoretical guarantees that one can achieve and tools of the trade. Towards the end of the talk, I will provide a brief overview of our recent results on hypergraph partitioning problems and their applications to computer vision and machine learning.
Bio:
Ambedkar Dukkipati received B. Tech from Department of Computer Science and Engineering, Indian Institute of Technology, Madras in 1999, Masters and PhD from Computer Science and Automation, Indian Institute of Science in 2002 and 2006, respectively. He was a postdoctoral researcher at EURANDOM (European Institute for Probability and Statistics) at Eindhoven before joining as a faculty in the department of Computer Science and Automation, Indian Institute of Science in 2009. His research interests are in statistical machine learning, spectral graph methods, algorithmic algebra.