Optimization of neural networks through grammatical evolution and a genetic algorithm

•Neural codification confers scalability and search space reduction.•The parallel genome scan engine increases the implicit parallelism of the GA.•Our approach rewards economical ANNs which have better generalization capacity.•Reduction in chromosome length from 512 to 180 bits.•Our NEA outperforms...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Expert systems with applications Ročník 56; s. 368 - 384
Hlavní autori: de Campos, Lídio Mauro Lima, de Oliveira, Roberto Célio Limão, Roisenberg, Mauro
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Elsevier Ltd 01.09.2016
Predmet:
Algorithms
Classification
Computer simulation
Construction
Evolutionary computation
Forecasting
Genetic algorithms
Grammatical evolution
Neural networks
Time series
Neural networks
Grammatical evolution
Evolutionary computation
ISSN:0957-4174, 1873-6793
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:•Neural codification confers scalability and search space reduction.•The parallel genome scan engine increases the implicit parallelism of the GA.•Our approach rewards economical ANNs which have better generalization capacity.•Reduction in chromosome length from 512 to 180 bits.•Our NEA outperforms other methods, providing the lowest computational effort. This paper proposes a hybrid neuro-evolutive algorithm (NEA) that uses a compact indirect encoding scheme (IES) for representing its genotypes (a set of ten production rules of a Lindenmayer System with memory), moreover has the ability to reuse the genotypes and automatically build modular, hierarchical and recurrent neural networks. A genetic algorithm (GA) evolves a Lindenmayer System (L-System) that is used to design the neural network’s architecture. This basic neural codification confers scalability and search space reduction in relation to other methods. Furthermore, the system uses a parallel genome scan engine that increases both the implicit parallelism and convergence of the GA. The fitness function of the NEA rewards economical artificial neural networks (ANNs) that are easily implemented. The NEA was tested on five real-world classification datasets and three well-known datasets for time series forecasting (TSF). The results are statistically compared against established state-of-the-art algorithms and various forecasting methods (ADANN, ARIMA, UCM, and Forecast Pro). In most cases, our NEA outperformed the other methods, delivering the most accurate classification and time series forecasting with the least computational effort. These superior results are attributed to the improved effectiveness and efficiency of NEA in the decision-making process. The result is an optimized neural network architecture for solving classification problems and simulating dynamical systems.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:0957-4174
1873-6793
DOI:10.1016/j.eswa.2016.03.012