Application of Reservoir Computing Based on a 2D Hyperchaotic Discrete Memristive Map in Efficient Temporal Signal Processing

Abstract

The analysis of time series is essential in many fields, and reservoir computing (RC) can provide effective temporal processing that makes it well-suited for time series analysis and prediction tasks. In this study, we introduce a new discrete memristor model and a corresponding two-dimensional hyperchaotic map with complex dynamic properties that are well-suited for reservoir computing. By applying this map to the RC, we enhance the state richness of the reservoir, resulting in improved performance. The paper evaluates the performance of the proposed RC approach using time series data for sunspot, exchange rate, and solar-E forecasting tasks. Our experimental results demonstrate that this approach is highly effective in handling temporal data with both accuracy and efficiency. And comparing with other discrete memristive chaotic maps, the proposed map is the best for improving the RC performance. Furthermore, the proposed RC model is characterized by a simple structure that enables it to fully exploit the time-dependence of the state values of the hyperchaotic map.