Design of Ensemble Neural Networks with Type-3 Fuzzy Aggregation using Particle Swarm Optimization and Genetic Algorithms for Ethereum Prediction

Martha Pulido, Patricia Melin, Oscar Castillo

Abstract


In this study, an ensemble neural network (ENN) for Ethereum time series prediction was optimized using particle swarm optimization and genetic algorithms. Additionally, Type-1, Type-2, and Type-3 fuzzy inference systems, of both Mamdani and Sugeno types, were designed for achieving the prediction. The integration performed with these fuzzy systems is achieved by utilizing the results from optimizing the ENN with each optimization algorithm. In this case, the Ethereum data is the series being used for testing the proposal. This approach aims to minimize prediction error by combining the responses of the ENN with Type-1, Type-2, and Type-3 fuzzy systems, each consisting of five inputs and consequently 32 fuzzy rules are utilized. The results show that the Type-1, Type-2, and Type-3 fuzzy system approach yields an accurate prediction of the Ethereum series, as further validated by statistical tests on the results of the fuzzy systems.

Keywords


Ethereum time series, type-3 fuzzy system, time series ensemble neural networks, mamdani model, sugeno model

Full Text: PDF