When viewing the Technical Program schedule, on the far righthand side
is a column labeled "PLANNER." Use this planner to build your own
schedule. Once you select an event and want to add it to your personal
schedule, just click on the calendar icon of your choice (outlook
calendar, ical calendar or google calendar) and that event will be
stored there. As you select events in this manner, you will have your
own schedule to guide you through the week.
You can also create your personal schedule on the SC11 app (Boopsie) on your smartphone. Simply select a session you want to attend and "add" it to your plan. Continue in this manner until you have created your own personal schedule. All your events will appear under "My Event Planner" on your smartphone.
Parallel Algorithms for Clustering and Nearest Neighbor Search in High Dimensions
SESSION: Research Poster Reception
EVENT TYPE: ACM Student Research Competition Poster, Poster, Electronic Poster
TIME: 5:15PM - 7:00PM
SESSION CHAIR: Bernd Mohr
AUTHOR(S):Logan Moon, Daniel Long, Shreyas Joshi, Vyomkesh Tripathi, Bo Xiao, George Biros
ROOM:WSCC North Galleria 2nd/3rd Floors
ABSTRACT: Although there is a significant body of work for efficient clustering and nearest neighbor searches in database systems, surprisingly little work has been done on algorithms and libraries for high-end scientific computing platforms
We present a set of tools implemented using MPI and OpenMP for massively parallel computational geometry problems. We combine parallel distance and filtering operations and kmeans clustering algorithms with an optimized seeding procedure, locality-sensitive hashing, and a novel parallel indexing structure, “SPRINKLE-TREE”, to support exact and approximate operations for problems in high dimension.
The overall scheme shows excellent scalability and enables the analysis of datasets of unprecedented scale. In our largest runs, we were able to conduct exact searches and clustering on a 45TB dataset in under 30 seconds on 98K cores on the Kraken platform. Tree construction takes under three seconds for a dataset with over 20 million points in 100 dimensions on 12288 cores.
Bernd Mohr (Chair) - Juelich Supercomputing Centre
Logan Moon - University of Texas at Austin
Daniel Long - Georgia Institute of Technology
Shreyas Joshi - Georgia Institute of Technology
Vyomkesh Tripathi - Georgia Institute of Technology