Nearest Neighbor Search Applications for the Terasys Massively Parallel Workstation.

Abstract

The Terasys workstation is a massively parallel computer developed at IDA's Center for Computing Sciences. This report, based on a project conducted in IDA's Computer and Software Engineering Division, presents an overview of the Terasys workstation and discusses how the Terasys could be applied to the task of nearest neighbor search. The report discusses a number of areas where Terasys based nearest neighbor search could potentially be applied including nearest neighbor classification, pattern recognition, and estimating intrinsic dimensionality. One area where the use of Terasys-based nearest neighbor search seems particularly promising is in the development of new neighbor classifiers. The report includes experimental results showing that a Terasys with 32,768 processors can perform a particular nearest neighbor search problem up to 69 times faster than a Sun Microsystems 61 MHz Sparcstation-20. To illustrate how the Terasys could be applied to the kinds of government work conducted by IDA divisions outside of the Center for Computing Sciences, the report examines three possible applications of Terasys-based nearest neighbor search to an ongoing radar evaluation project in IDA's Science and Technology Division.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 1996
Accession Number
ADA318137

Entities

People

  • Eric W. Johnson

Organizations

  • Institute for Defense Analyses

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Algorithms
  • Classification
  • Computations
  • Computer Programming
  • Computer Science
  • Computers
  • Dimensionality Reduction
  • Engineering
  • Floating Point Operations
  • Host Computers
  • Machine Learning
  • Network Science
  • Pattern Recognition
  • Radar
  • Random Variables
  • Recognition
  • Software Development

Fields of Study

  • Computer science

Readers

  • Computer Vision.
  • Defense Technology Research and Development.
  • Neurodegenerative Parkinson's Disease and Rickettsial Disease handbook, including the data level of dopamine, BC, neurons, and PD.

Technology Areas

  • AI & ML