Academic Bibliography

 Search 200 years of publications by Ghent University researchers.
Advanced search
1 file | 3.36 MB
Add to list
  1. Search results
  2. Investigating parallel implementation...

Investigating parallel implementations of genetic algorithms for stochastic part-of-speech tagging

Shimanto Rahman (UGent) , Matthias Bogaert (UGent) and Dirk Van den Poel (UGent)
(2022) EURO 2022 : conference handbook and abstracts : 32nd European Conference on Operational Research (EURO XXXII). p.176-176
Author
Organization
Abstract
This paper aims to provide a comprehensive benchmark of the performance and speed of parallel architectures of genetic algorithms compared to serial architectures. In particular, four different popular configurations of genetic algorithms are discussed: a panmictic, a distributed, a circularly linked cellular, and a toroidally linked cellular design. The genetic algorithms are evaluated on a variety of popular corpora (i.e., Penn Treebank, Brown, and Susanne) in stochastic part-of-speech tagging. Preliminary results show that on the universal tag set the cellular genetic algorithms achieve superior performance in token, unknown words, and sentence accuracy. Particularly the circularly linked cellular genetic algorithms converge quicker to a (local) optimum. Parallel genetic algorithms are considerably more robust against unknown words compared to more traditional optimization algorithms used in Hidden Markov Models. Compared to previous benchmark studies in the same area, our study examines a wider range of data sets, a more comprehensive set of performance measures, and a more extensive set of parallel architectures.
Keywords
Data Science, Natural Language Processing

Downloads

  • Download
    (...).pdf
    • full text (Published version)
    • |
    • UGent only
    • |
    • PDF
    • |
    • 3.36 MB

Citation

Please use this url to cite or link to this publication:

MLA
Rahman, Shimanto, et al. “Investigating Parallel Implementations of Genetic Algorithms for Stochastic Part-of-Speech Tagging.” EURO 2022 : Conference Handbook and Abstracts : 32nd European Conference on Operational Research (EURO XXXII), Association of European Operational Research Societies, 2022, pp. 176–176.
APA
Rahman, S., Bogaert, M., & Van den Poel, D. (2022). Investigating parallel implementations of genetic algorithms for stochastic part-of-speech tagging. EURO 2022 : Conference Handbook and Abstracts : 32nd European Conference on Operational Research (EURO XXXII), 176–176. Espoo, Finland: Association of European Operational Research Societies.
Chicago author-date
Rahman, Shimanto, Matthias Bogaert, and Dirk Van den Poel. 2022. “Investigating Parallel Implementations of Genetic Algorithms for Stochastic Part-of-Speech Tagging.” In EURO 2022 : Conference Handbook and Abstracts : 32nd European Conference on Operational Research (EURO XXXII), 176–176. Espoo, Finland: Association of European Operational Research Societies.
Chicago author-date (all authors)
Rahman, Shimanto, Matthias Bogaert, and Dirk Van den Poel. 2022. “Investigating Parallel Implementations of Genetic Algorithms for Stochastic Part-of-Speech Tagging.” In EURO 2022 : Conference Handbook and Abstracts : 32nd European Conference on Operational Research (EURO XXXII), 176–176. Espoo, Finland: Association of European Operational Research Societies.
Vancouver
1.
Rahman S, Bogaert M, Van den Poel D. Investigating parallel implementations of genetic algorithms for stochastic part-of-speech tagging. In: EURO 2022 : conference handbook and abstracts : 32nd European Conference on Operational Research (EURO XXXII). Espoo, Finland: Association of European Operational Research Societies; 2022. p. 176–176.
IEEE
[1]
S. Rahman, M. Bogaert, and D. Van den Poel, “Investigating parallel implementations of genetic algorithms for stochastic part-of-speech tagging,” in EURO 2022 : conference handbook and abstracts : 32nd European Conference on Operational Research (EURO XXXII), Espoo, Finland, 2022, pp. 176–176.
@inproceedings{01HR48VV73WWRTMGXNCM7S09QC,
  abstract     = {{This paper aims to provide a comprehensive benchmark of the performance and speed of parallel architectures of genetic algorithms compared to serial architectures. In particular, four different popular configurations of genetic algorithms are discussed: a panmictic, a distributed, a circularly linked cellular, and a toroidally linked cellular design. The genetic algorithms are evaluated on a variety of popular corpora (i.e., Penn Treebank, Brown, and Susanne) in stochastic part-of-speech tagging. Preliminary results show that on the universal tag set the cellular genetic algorithms achieve superior performance in token, unknown words, and sentence accuracy. Particularly the circularly linked cellular genetic algorithms converge quicker to a (local) optimum. Parallel genetic algorithms are considerably more robust against unknown words compared to more traditional optimization algorithms used in Hidden Markov Models. Compared to previous benchmark studies in the same area, our study examines a wider range of data sets, a more comprehensive set of performance measures, and a more extensive set of parallel architectures.}},
  author       = {{Rahman, Shimanto and Bogaert, Matthias and Van den Poel, Dirk}},
  booktitle    = {{EURO 2022 : conference handbook and abstracts : 32nd European Conference on Operational Research (EURO XXXII)}},
  isbn         = {{9789519525419}},
  keywords     = {{Data Science,Natural Language Processing}},
  language     = {{eng}},
  location     = {{Espoo, Finland}},
  pages        = {{176--176}},
  publisher    = {{Association of European Operational Research Societies}},
  title        = {{Investigating parallel implementations of genetic algorithms for stochastic part-of-speech tagging}},
  url          = {{https://euro2022espoo.com/}},
  year         = {{2022}},
}