Random walks

Tuesday, October 29, 2013 - 2:00pm - 2:50pm
Ali Jadbabaie (University of Pennsylvania)
In this talk I will show how tools from algebraic topology, specifically homology and cohomology theory, can be used to study a range of problems in sensor networks from distributed, location-free coverage verification to distributed estimation based on pairwise relative measurements. I will show that finding sparse homology generators correspond to location of coverage holes, even when coordinates are not available.
Tuesday, October 29, 2013 - 10:15am - 11:05am
Jie Gao (State University of New York, Stony Brook (SUNY))
Random walk on a graph is a Markov chain and thus is ‘memoryless’ as the next node to visit depends only on the current node and not on the sequence of events that
Friday, November 1, 2013 - 9:00am - 9:50am
Sayan Mukherjee (Duke University)
I will discuss random walks on simplicial complexes, Cheeger inequalities,
and mixing times. The talk will summarize the results of two papers that explore
how random walks on walks graphs extend to simplicial complexes. I will also discus two papers that study higher order Laplacians and expander properties.
Two possible applications in machine learning will be discussed: spectral clustering and label (edge) propogation.

joint work with John Steenbergen
Subscribe to RSS - Random walks