Parallel and distributed systems for constructive neural network learning

Justin Fletcher, Z. Obradovic

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

A constructive learning algorithm dynamically creates a problem-specific neural network architecture rather than learning on a pre-specified architecture. The authors propose a parallel version of their recently presented constructive neural network learning algorithm. Parallelization provides a computational speedup by a factor of O(t) where t is the number of training examples. Distributed and parallel implementations under p4 using a network of workstations and a Touchstone DELTA are examined. Experimental results indicate that algorithm parallelization may result not only in improved computational time, but also in better prediction quality.

Original languageEnglish (US)
Title of host publicationProceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages174-178
Number of pages5
ISBN (Electronic)0818639008
DOIs
StatePublished - Jan 1 1993
Externally publishedYes
Event2nd International Symposium on High Performance Distributed Computing, HPDC 1993 - Spokane, United States
Duration: Jul 20 1993Jul 23 1993

Publication series

NameProceedings of the IEEE International Symposium on High Performance Distributed Computing
ISSN (Print)1082-8907

Conference

Conference2nd International Symposium on High Performance Distributed Computing, HPDC 1993
CountryUnited States
CitySpokane
Period7/20/937/23/93

Fingerprint

Learning algorithms
Neural networks
Network architecture

ASJC Scopus subject areas

  • Software
  • Computer Networks and Communications

Cite this

Fletcher, J., & Obradovic, Z. (1993). Parallel and distributed systems for constructive neural network learning. In Proceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993 (pp. 174-178). [263844] (Proceedings of the IEEE International Symposium on High Performance Distributed Computing). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/HPDC.1993.263844

Parallel and distributed systems for constructive neural network learning. / Fletcher, Justin; Obradovic, Z.

Proceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993. Institute of Electrical and Electronics Engineers Inc., 1993. p. 174-178 263844 (Proceedings of the IEEE International Symposium on High Performance Distributed Computing).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Fletcher, J & Obradovic, Z 1993, Parallel and distributed systems for constructive neural network learning. in Proceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993., 263844, Proceedings of the IEEE International Symposium on High Performance Distributed Computing, Institute of Electrical and Electronics Engineers Inc., pp. 174-178, 2nd International Symposium on High Performance Distributed Computing, HPDC 1993, Spokane, United States, 7/20/93. https://doi.org/10.1109/HPDC.1993.263844
Fletcher J, Obradovic Z. Parallel and distributed systems for constructive neural network learning. In Proceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993. Institute of Electrical and Electronics Engineers Inc. 1993. p. 174-178. 263844. (Proceedings of the IEEE International Symposium on High Performance Distributed Computing). https://doi.org/10.1109/HPDC.1993.263844
Fletcher, Justin ; Obradovic, Z. / Parallel and distributed systems for constructive neural network learning. Proceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993. Institute of Electrical and Electronics Engineers Inc., 1993. pp. 174-178 (Proceedings of the IEEE International Symposium on High Performance Distributed Computing).
@inproceedings{be1a678416644b75b5762af7f7ba5587,
title = "Parallel and distributed systems for constructive neural network learning",
abstract = "A constructive learning algorithm dynamically creates a problem-specific neural network architecture rather than learning on a pre-specified architecture. The authors propose a parallel version of their recently presented constructive neural network learning algorithm. Parallelization provides a computational speedup by a factor of O(t) where t is the number of training examples. Distributed and parallel implementations under p4 using a network of workstations and a Touchstone DELTA are examined. Experimental results indicate that algorithm parallelization may result not only in improved computational time, but also in better prediction quality.",
author = "Justin Fletcher and Z. Obradovic",
year = "1993",
month = "1",
day = "1",
doi = "10.1109/HPDC.1993.263844",
language = "English (US)",
series = "Proceedings of the IEEE International Symposium on High Performance Distributed Computing",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "174--178",
booktitle = "Proceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993",

}

TY - GEN

T1 - Parallel and distributed systems for constructive neural network learning

AU - Fletcher, Justin

AU - Obradovic, Z.

PY - 1993/1/1

Y1 - 1993/1/1

N2 - A constructive learning algorithm dynamically creates a problem-specific neural network architecture rather than learning on a pre-specified architecture. The authors propose a parallel version of their recently presented constructive neural network learning algorithm. Parallelization provides a computational speedup by a factor of O(t) where t is the number of training examples. Distributed and parallel implementations under p4 using a network of workstations and a Touchstone DELTA are examined. Experimental results indicate that algorithm parallelization may result not only in improved computational time, but also in better prediction quality.

AB - A constructive learning algorithm dynamically creates a problem-specific neural network architecture rather than learning on a pre-specified architecture. The authors propose a parallel version of their recently presented constructive neural network learning algorithm. Parallelization provides a computational speedup by a factor of O(t) where t is the number of training examples. Distributed and parallel implementations under p4 using a network of workstations and a Touchstone DELTA are examined. Experimental results indicate that algorithm parallelization may result not only in improved computational time, but also in better prediction quality.

UR - http://www.scopus.com/inward/record.url?scp=84953478623&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84953478623&partnerID=8YFLogxK

U2 - 10.1109/HPDC.1993.263844

DO - 10.1109/HPDC.1993.263844

M3 - Conference contribution

T3 - Proceedings of the IEEE International Symposium on High Performance Distributed Computing

SP - 174

EP - 178

BT - Proceedings of the 2nd International Symposium on High Performance Distributed Computing, HPDC 1993

PB - Institute of Electrical and Electronics Engineers Inc.

ER -