Evolutionary Approximation and Neural Architecture Search

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26230%2F22%3APU145799" target="_blank" >RIV/00216305:26230/22:PU145799 - isvavai.cz</a>

Výsledek na webu

<a href="https://link.springer.com/article/10.1007/s10710-022-09441-z" target="_blank" >https://link.springer.com/article/10.1007/s10710-022-09441-z</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/s10710-022-09441-z" target="_blank" >10.1007/s10710-022-09441-z</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Evolutionary Approximation and Neural Architecture Search

Popis výsledku v původním jazyce

Automated neural architecture search (NAS) methods are now employed to routinely deliver high-quality neural network architectures for various challenging data sets and reduce the designers effort. The NAS methods utilizing multi-objective evolutionary algorithms are especially useful when the objective is not only to minimize the network error but also to reduce the number of parameters (weights) or power consumption of the inference phase. We propose a multi-objective NAS method based on Cartesian genetic programming for evolving convolutional neural networks (CNN). The method allows approximate operations to be used in CNNs to reduce the power consumption of a target hardware implementation. During the NAS process, a suitable CNN architecture is evolved together with selecting approximate multipliers to deliver the best trade-offs between accuracy, network size, and power consumption. The most suitable 8 x N-bit approximate multipliers are automatically selected from a library of approximate multipliers. Evolved CNNs are compared with CNNs developed by other NAS methods on the CIFAR-10 and SVHN benchmark problems.

Název v anglickém jazyce

Evolutionary Approximation and Neural Architecture Search

Popis výsledku anglicky

Automated neural architecture search (NAS) methods are now employed to routinely deliver high-quality neural network architectures for various challenging data sets and reduce the designers effort. The NAS methods utilizing multi-objective evolutionary algorithms are especially useful when the objective is not only to minimize the network error but also to reduce the number of parameters (weights) or power consumption of the inference phase. We propose a multi-objective NAS method based on Cartesian genetic programming for evolving convolutional neural networks (CNN). The method allows approximate operations to be used in CNNs to reduce the power consumption of a target hardware implementation. During the NAS process, a suitable CNN architecture is evolved together with selecting approximate multipliers to deliver the best trade-offs between accuracy, network size, and power consumption. The most suitable 8 x N-bit approximate multipliers are automatically selected from a library of approximate multipliers. Evolved CNNs are compared with CNNs developed by other NAS methods on the CIFAR-10 and SVHN benchmark problems.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/GA21-13001S" target="_blank" >GA21-13001S: Automatizovaný návrh hardwarových akcelerátorů pro strojového učení zohledňující výpočetní zdroje</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2022

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Genetic Programming and Evolvable Machines

ISSN

1389-2576

e-ISSN

1573-7632

Svazek periodika

23

Číslo periodika v rámci svazku

3

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

24

Strana od-do

351-374

Kód UT WoS článku

000810226500001

EID výsledku v databázi Scopus

2-s2.0-85131746167