Large datasets: a mixed method to adapt and improve their learning by neural networks used in regression contexts - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2011

Large datasets: a mixed method to adapt and improve their learning by neural networks used in regression contexts

Abstract

The purpose of this work is to further study the relevance of accelerating the Monte-Carlo calculations for the gamma rays external radiotherapy through feed-forward neural networks. We have previously presented a parallel incremental algorithm that builds neural networks of reduced size, while providing high quality approximations of the dose deposit~\cite{Vecpar08b}. Our parallel algorithm consists in an optimized decomposition of the initial learning dataset (also called learning domain) in as much subsets as available processors. However, although that decomposition provides subsets of similar signal complexities, their sizes may be quite different, still implying potential differences in their learning times. This paper presents an efficient data extraction allowing a good and balanced training without any loss of signal information. As will be shown, the resulting irregular decomposition permits an important improvement in the learning time of the global network.
Fichier principal
Vignette du fichier
978-3-642-23957-1_21_Chapter.pdf (777.64 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-00643870 , version 1 (02-08-2017)

Licence

Attribution

Identifiers

Cite

Marc Sauget, Julien Henriet, Michel Salomon, Sylvain Contassot-Vivier. Large datasets: a mixed method to adapt and improve their learning by neural networks used in regression contexts. 12th Engineering Applications of Neural Networks (EANN 2011) and 7th Artificial Intelligence Applications and Innovations (AIAI), Sep 2011, Corfu, Greece. pp.182-191, ⟨10.1007/978-3-642-23957-1_21⟩. ⟨hal-00643870⟩
649 View
72 Download

Altmetric

Share

Gmail Facebook X LinkedIn More