Parallel Processing of Compressed Graphs with Ligra+

Calendar
When:
December 7, 2017 @ 5:00 pm – 6:30 pm America/New York Timezone
2017-12-07T17:00:00-05:00
2017-12-07T18:30:00-05:00
Where:
Boston University - Room EPC 207
All Meetings
Nuclear & Plasma Sciences

Nuclear and Plasma Science Society

A seminar sponsored by IEEE Boston Chapter of Nuclear and Plasma Science Society (NPSS) and Boston University IEEE HKN

Parallel Processing of Compressed Graphs with Ligra+

5:00 – 6:30 PM

Speaker: Katie Lewis, MIT

Graph algorithms span many applications from biological network to social network analysis. With the continuous growth in data, compression has become increasingly important for reducing the space and cost of large-scale graph analytics. In this talk, I will give an overview of several integer compression techniques, such as StreamVbyte, currently used in the literature. I will also discuss how to apply these techniques towards compressing graphs and speeding up algorithms in the graph processing framework, Ligra+.

Katie Lewis is a graduate student conducting research in Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT. Her research with Professor Julian Shun focuses on the theory and implementation of parallel graph processing frameworks and algorithms. In particular, her current work incorporates integer compression techniques into the graph processing framework Ligra+ to reduce the space and computation requirements of large-scale graph processing. When Katie Lewis studied at Boston University for a BS in Computer Engineering, she served as President of BU IEEE/IEEE-HKN. She is elected as a student governor to the board of IEEE HKN for 2018.”

Meeting Location: Boston University, Room EPC 207, 750 Commonwealth Avenue, Boston, MA 02215.