On the Optimization of Machine Learning Techniques for Chaotic Time Series Prediction

Interest in chaotic time series prediction has grown recent years due to its multiple applications fields such as climate and health. In this work, we summarize the contribution of works that use different machine learning (ML) methods predict series. It is highlighted challenge predicting larger horizon with low error, for task, majority authors datasets generated by systems Lorenz, Rössler Mackey–Glass. Among classification description methods, work takes a case study Echo State Network (ESN) show optimization can lead enhance Different applied ones are given appreciate metaheuristics good option optimize an ESN. manner, ESN closed-loop mode optimized herein applying Particle Swarm Optimization. The results increase about twice number steps ahead, thus highlighting usefulness performing hyperparameters ML method horizon.