Abstract
Analyzing relationships among objects is essential for variety tasks of statistical data analysis. Graph is an established way to describe those relationships mathematically, in which node represents some object and edge connecting two nodes represents existence of a relationship between two nodes. For instance, protein-protein interaction network data has attracted wide interest in the field of bioinformatics to understand underlying biological systems, in which we can regard a protein as a node and an interaction as an edge in a graph. A field called machine learning has developed many general techniques to analyze data represented as a graph. In this talk, I'll present basic ideas and some of recent techniques of graph analysis in machine learning.
