A pseudo-softmax function for hardware-based high speed image classification

Abstract In this work a novel architecture, named pseudo-softmax, to compute an approximated form of the softmax function is presented. This architecture can be fruitfully used in the last layer of Neural Networks and Convolutional Neural Networks for classification tasks, and in Reinforcement Learn...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Gian Carlo Cardarilli, Luca Di Nunzio, Rocco Fazzolari, Daniele Giardino, Alberto Nannarelli, Marco Re, Sergio Spanò
Formato:	article
Lenguaje:	EN
Publicado:	Nature Portfolio 2021
Materias:	Medicine R Science Q
Acceso en línea:	https://doaj.org/article/7fd239c516c54de8b74b325a8777bbfd
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:7fd239c516c54de8b74b325a8777bbfd
record_format	dspace
spelling	oai:doaj.org-article:7fd239c516c54de8b74b325a8777bbfd2021-12-02T18:46:55ZA pseudo-softmax function for hardware-based high speed image classification10.1038/s41598-021-94691-72045-2322https://doaj.org/article/7fd239c516c54de8b74b325a8777bbfd2021-07-01T00:00:00Zhttps://doi.org/10.1038/s41598-021-94691-7https://doaj.org/toc/2045-2322Abstract In this work a novel architecture, named pseudo-softmax, to compute an approximated form of the softmax function is presented. This architecture can be fruitfully used in the last layer of Neural Networks and Convolutional Neural Networks for classification tasks, and in Reinforcement Learning hardware accelerators to compute the Boltzmann action-selection policy. The proposed pseudo-softmax design, intended for efficient hardware implementation, exploits the typical integer quantization of hardware-based Neural Networks obtaining an accurate approximation of the result. In the paper, a detailed description of the architecture is given and an extensive analysis of the approximation error is performed by using both custom stimuli and real-world Convolutional Neural Networks inputs. The implementation results, based on CMOS standard-cell technology, compared to state-of-the-art architectures show reduced approximation errors.Gian Carlo CardarilliLuca Di NunzioRocco FazzolariDaniele GiardinoAlberto NannarelliMarco ReSergio SpanòNature PortfolioarticleMedicineRScienceQENScientific Reports, Vol 11, Iss 1, Pp 1-10 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Medicine R Science Q
spellingShingle	Medicine R Science Q Gian Carlo Cardarilli Luca Di Nunzio Rocco Fazzolari Daniele Giardino Alberto Nannarelli Marco Re Sergio Spanò A pseudo-softmax function for hardware-based high speed image classification
description	Abstract In this work a novel architecture, named pseudo-softmax, to compute an approximated form of the softmax function is presented. This architecture can be fruitfully used in the last layer of Neural Networks and Convolutional Neural Networks for classification tasks, and in Reinforcement Learning hardware accelerators to compute the Boltzmann action-selection policy. The proposed pseudo-softmax design, intended for efficient hardware implementation, exploits the typical integer quantization of hardware-based Neural Networks obtaining an accurate approximation of the result. In the paper, a detailed description of the architecture is given and an extensive analysis of the approximation error is performed by using both custom stimuli and real-world Convolutional Neural Networks inputs. The implementation results, based on CMOS standard-cell technology, compared to state-of-the-art architectures show reduced approximation errors.
format	article
author	Gian Carlo Cardarilli Luca Di Nunzio Rocco Fazzolari Daniele Giardino Alberto Nannarelli Marco Re Sergio Spanò
author_facet	Gian Carlo Cardarilli Luca Di Nunzio Rocco Fazzolari Daniele Giardino Alberto Nannarelli Marco Re Sergio Spanò
author_sort	Gian Carlo Cardarilli
title	A pseudo-softmax function for hardware-based high speed image classification
title_short	A pseudo-softmax function for hardware-based high speed image classification
title_full	A pseudo-softmax function for hardware-based high speed image classification
title_fullStr	A pseudo-softmax function for hardware-based high speed image classification
title_full_unstemmed	A pseudo-softmax function for hardware-based high speed image classification
title_sort	pseudo-softmax function for hardware-based high speed image classification
publisher	Nature Portfolio
publishDate	2021
url	https://doaj.org/article/7fd239c516c54de8b74b325a8777bbfd
work_keys_str_mv	AT giancarlocardarilli apseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT lucadinunzio apseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT roccofazzolari apseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT danielegiardino apseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT albertonannarelli apseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT marcore apseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT sergiospano apseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT giancarlocardarilli pseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT lucadinunzio pseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT roccofazzolari pseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT danielegiardino pseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT albertonannarelli pseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT marcore pseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification AT sergiospano pseudosoftmaxfunctionforhardwarebasedhighspeedimageclassification
_version_	1718377687956848640

A pseudo-softmax function for hardware-based high speed image classification

Ejemplares similares