Designing neural networks for object recognition requires considerable architecture engineering. As a remedy, neuro-evolutionary network architecture search, which automatically searches for optimal network architectures using evolutionary algorithms, has recently become very popular. Although very effective, evolutionary algorithms rely heavily on having a large population of individuals (i.e., network architectures) and is therefore memory expensive. In this work, we propose a Regularized Evolutionary Algorithm with low memory footprint to evolve a dynamic image classifier. In details, we introduce novel custom operators that regularize the evolutionary process of a micro-population of 10 individuals. We conduct experiments on three different digits datasets (MNIST, USPS, SVHN) and show that our evolutionary method obtains competitive results with the current state-of-the-art.

Regularized evolutionary algorithm for dynamic neural topology search / Saltori, Cristiano; Roy, S.; Sebe, N.; Iacca, G.. - 11751:(2019), pp. 219-230. (Intervento presentato al convegno International Conference on Image Analysis and Processing tenutosi a Trento nel 9-13 September, 2019) [10.1007/978-3-030-30642-7_20].

Regularized evolutionary algorithm for dynamic neural topology search

Saltori, Cristiano;S. Roy;N. Sebe;G. Iacca
2019-01-01

Abstract

Designing neural networks for object recognition requires considerable architecture engineering. As a remedy, neuro-evolutionary network architecture search, which automatically searches for optimal network architectures using evolutionary algorithms, has recently become very popular. Although very effective, evolutionary algorithms rely heavily on having a large population of individuals (i.e., network architectures) and is therefore memory expensive. In this work, we propose a Regularized Evolutionary Algorithm with low memory footprint to evolve a dynamic image classifier. In details, we introduce novel custom operators that regularize the evolutionary process of a micro-population of 10 individuals. We conduct experiments on three different digits datasets (MNIST, USPS, SVHN) and show that our evolutionary method obtains competitive results with the current state-of-the-art.
2019
International Conference on Image Analysis and Processing
Heidelberg
Springer
Saltori, Cristiano; Roy, S.; Sebe, N.; Iacca, G.
Regularized evolutionary algorithm for dynamic neural topology search / Saltori, Cristiano; Roy, S.; Sebe, N.; Iacca, G.. - 11751:(2019), pp. 219-230. (Intervento presentato al convegno International Conference on Image Analysis and Processing tenutosi a Trento nel 9-13 September, 2019) [10.1007/978-3-030-30642-7_20].
File in questo prodotto:
File Dimensione Formato  
178 (2).pdf

Solo gestori archivio

Tipologia: Versione editoriale (Publisher’s layout)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 434.76 kB
Formato Adobe PDF
434.76 kB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11572/250688
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 9
  • ???jsp.display-item.citation.isi??? 8
social impact