BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20101116T231500Z
DTEND:20101117T010000Z
LOCATION:Main Lobby
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Counting the number of template embeddings in large network has been found to be useful in a number of applications, such as biological and social networks. However, due to large computational cost, template counting remains a challenging problem, and all prior results have considered networks=0Awith a few thousand nodes. =0A=0AWe develop a parallel subgraph counting algorithm, ParSE, that scales to networks with millions of nodes. Our algorithm is a randomized approximation scheme, that estimates the subgraph frequency to any desired level of accuracy, and allows enumeration of a class of motifs that extends the tree-like templates which is considered in prior work. Our approach is based on parallelization of so called color coding technique, combined with a stream based graph partitioning.  =0A=0AThe detailed algorithm, implementation issues, as well as experiment results will be presented in the poster session.
SUMMARY:Parallel Color Coding and Graph Partitioning Enabling Subgraph Counting for Massive Graphs
PRIORITY:3
END:VEVENT
END:VCALENDAR