Lightweight CNN Frameworks and their Optimization using Evolutionary Algorithms

Yangyang Chang, Gerald E. Sobelman

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

Abstract

This paper presents novel frameworks for efficient and lightweight convolutional neural networks that are suitable for use in embedded system applications. Population-based metaheuristic approaches including the genetic algorithm, cuckoo search, the multifactorial evolutionary algorithm and a proposed hybrid approach are used to optimize their performance on image classification tasks. The methods utilize small population sizes without requiring weight-sharing or a surrogate function, and both binary and integer encoding methods are applied in the optimization. The results from these various strategies are evaluated using metrics of computation time and classification accuracy. The multifactorial-based approach is found to give the highest classification accuracy and it requires only a moderate evaluation time. Also, comparisons with prior approaches demonstrate that these methods show a favorable tradeoff between accuracy and computational cost.

Original languageEnglish (US)
Title of host publicationProceedings of the 2022 International Electrical Engineering Congress, iEECON 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665402064
DOIs
StatePublished - 2022
Event2022 International Electrical Engineering Congress, iEECON 2022 - Khon Kaen, Thailand
Duration: Mar 9 2022Mar 11 2022

Publication series

NameProceedings of the 2022 International Electrical Engineering Congress, iEECON 2022

Conference

Conference2022 International Electrical Engineering Congress, iEECON 2022
Country/TerritoryThailand
CityKhon Kaen
Period3/9/223/11/22

Bibliographical note

Publisher Copyright:
© 2022 IEEE.

Keywords

  • convolutional neural networks
  • cuckoo search
  • genetic algorithm
  • multifactorial evolutionary algorithm
  • network architecture search

Fingerprint

Dive into the research topics of 'Lightweight CNN Frameworks and their Optimization using Evolutionary Algorithms'. Together they form a unique fingerprint.

Cite this