Abstract
Random Walks on Graphs: Large-time Behavior and Applications to Analysis of Large Data Sets
Mauro Maggioni
Johns Hopkins University
We will discuss random walks on graphs, and how the corresponding eigenfunctions and heat kernels can be used to study the geometry, locally and globally, of such graphs. We show applications to the analysis of high dimensional data sets, that have an intrinsic geometric structure that can be modeled by graphs, and show how the above ideas can be applied to dimensionality reduction and machine learning tasks.
No video available