Kanerva's sparse distributed memory: an object-oriented implementation on the connection machine

  • Authors:
  • Andreas Turk;Gunther Gorz

  • Affiliations:
  • German Research Center for Artificial Intelligence, Kaiserslautern, Germany and Univ. of Erlangen-Nürnberg;Univ. of Erlangen-Nürnberg, IMMD, Erlangen, Germany

  • Venue:
  • IJCAI'95 Proceedings of the 14th international joint conference on Artificial intelligence - Volume 1
  • Year:
  • 1995

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper reports on an implementation of Kanerva's Sparse Distributed Memory for the Connection Machine. In order to accomplish a modular and adaptive software library we applied a plain object-oriented programming style to the Common Lisp extension ltsp. Some variations of the original model, the selected coordinate design, the hyperplane design, and a new general design, as well as the folded SDM due to Kanerva are realized. It has been necessary to elaborate a uniform presentation of the theoretical foundations the different designs are based on. We demonstrate the simulator's functionality with some simple applications. Runtime comparisons are given. We encourage the use of our simulation tool when outlining research topics of special interest to SDM.