Short Course

Morteza Alamgir, Department of Computer Science, University of Hamburg
Supervised and unsupervised learning with graphs

Many machine learning algorithms use graphs to model relations between data points. In my talk, i would discuss elementary properties of such graphs such as distances between graph vertices. Such properties can completely change depending on how we construct the graph. We would see the kind of information that we may lose when we transform our data to a graph. Finally, I would show that graph-based techniques can give surprising solutions to problems that a priori don’t involve graphs at all.

9:30-10:45 Introduction: Graphs and their applications in machine learning

11:15-12:30 Commute time, p-resistance distance and their limit behavior

13:30-15:00 Shortest path distance in random geometric graphs

Date & Time: 
Monday, 14 September, 2015, 9:30 - 15:00
Room 221, Department of Mathematical Sciences, Sharif University of Technology