Competition and Collaboration in Cooperative Coevolution of Elman Recurrent Neural Networks for Time-Series Prediction

Abstract

Collaboration enables weak species to survive in an environment where different species compete for limited resources. Cooperative coevolution (CC) is a nature-inspired optimization method that divides a problem into subcomponents and evolves them while genetically isolating them. Problem decomposition is an important aspect in using CC for neuroevolution. CC employs different problem decomposition methods to decompose the neural network training problem into subcomponents. Different problem decomposition methods have features that are helpful at different stages in the evolutionary process. Adaptation, collaboration, and competition are needed for CC, as multiple subpopulations are used to represent the problem. It is important to add collaboration and competition in CC. This paper presents a competitive CC method for training recurrent neural networks for chaotic time-series prediction. Two different instances of the competitive method are proposed that employs different problem decomposition methods to enforce island-based competition. The results show improvement in the performance of the proposed methods in most cases when compared with standalone CC and other methods from the literature.

Time Series Prediction With a Hybrid System Formed by Artificial Neural Network and Cognitive Development Optimization Algorithm

“… Chandra has used a Recurrent Neural Networks system trained thanks to the technique of Cooperative Coevolution (CC) to predict chaotic time series [45].…”

Using a Single Dendritic Neuron to Forecast Tourist Arrivals to Japan

“…The time-delayed ANN may be the simplest choice for representing a wide range of mappings between past and present values [38], but the fixed time delays in these ANNs remain constant throughout training after initialization, thereby risking a mismatch between the choice of time delay values and the temporal locations of important information in the input patterns [39]. The Elman recurrent ANN [40] has advantages compared with the MLP because the memory features obtained using a feedback mechanism can be used to extract time dependencies from the data. However, the traditional recurrent ANN algorithms based on the gradient descent approach are well known for their slow convergence and high computational costs [41], thus it is difficult to utilize them in actual applications.…”

Wind speed forecasting approach using secondary decomposition algorithm and Elman neural networks

“…The Elman neural network is a well-known recurrent network, which was proposed by scientist Elman in 1990 [33]. Comparing to other feed-forward neural networks, the Elman neural networks have additional recurrent layers.…”