VERSATILE FUNCTION IN GPA

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216275%3A25530%2F21%3A39917356" target="_blank" >RIV/00216275:25530/21:39917356 - isvavai.cz</a>

Výsledek na webu

<a href="http://www.nnw.cz" target="_blank" >http://www.nnw.cz</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.14311/NNW.2020.30.025" target="_blank" >10.14311/NNW.2020.30.025</a>

Jazyk výsledku

angličtina

Název v původním jazyce

VERSATILE FUNCTION IN GPA

Popis výsledku v původním jazyce

The paper, devoted to continuous versatile function application in the Genetic Programming Algorithm (GPA), begins with a discussion of similarities between GPA with versatile function and neural network. Then, the function set influence on GPA efficiency is discussed. In the next part, there is described a hybrid evolutionary algorithm that combines GPA for structure development and Evolutionary Strategy (ES) for parameters and constant optimization; which is herein much more significant than in the standard GPA. There is also discussed the setting of parameters of this hybrid algorithm and due to a different function set. The original idea of a versatile function, which origins come from the area of fuzzy control systems, is formulated and explained. Three different implementations of this versatile function are discussed. On the base of experiments with the hybrid evolutionary algorithm providing symbolic regression task of precomputed Lorenz attractor system data representing its dynamic behaviour; the comparison of three variants of versatile functions was formulated. The paper also presents ways how to set up hybrid evolutionary algorithm parameters like population sizes as well as limits of maximal population numbers for both algorithms: GPA for structural development and nested ES for parameters optimization. The versatile function concept is applicable but it requires the hybrid evolutionary algorithm use as it is explained in the paper.

Název v anglickém jazyce

VERSATILE FUNCTION IN GPA

Popis výsledku anglicky

The paper, devoted to continuous versatile function application in the Genetic Programming Algorithm (GPA), begins with a discussion of similarities between GPA with versatile function and neural network. Then, the function set influence on GPA efficiency is discussed. In the next part, there is described a hybrid evolutionary algorithm that combines GPA for structure development and Evolutionary Strategy (ES) for parameters and constant optimization; which is herein much more significant than in the standard GPA. There is also discussed the setting of parameters of this hybrid algorithm and due to a different function set. The original idea of a versatile function, which origins come from the area of fuzzy control systems, is formulated and explained. Three different implementations of this versatile function are discussed. On the base of experiments with the hybrid evolutionary algorithm providing symbolic regression task of precomputed Lorenz attractor system data representing its dynamic behaviour; the comparison of three variants of versatile functions was formulated. The paper also presents ways how to set up hybrid evolutionary algorithm parameters like population sizes as well as limits of maximal population numbers for both algorithms: GPA for structural development and nested ES for parameters optimization. The versatile function concept is applicable but it requires the hybrid evolutionary algorithm use as it is explained in the paper.

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/EF17_049%2F0008394" target="_blank" >EF17_049/0008394: Spolupráce Univerzity Pardubice a aplikační sféry v aplikačně orientovaném výzkumu lokačních, detekčních a simulačních systémů pro dopravní a přepravní procesy (PosiTrans)</a><br>

Návaznosti

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

Rok uplatnění

2021

Kód důvěrnosti údajů

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

Název periodika

Neural Network World

ISSN

1210-0552

e-ISSN

—

Svazek periodika

30

Číslo periodika v rámci svazku

6

Stát vydavatele periodika

CZ - Česká republika

Počet stran výsledku

14

Strana od-do

379-392

Kód UT WoS článku

000625072400003

EID výsledku v databázi Scopus

—