Echo State Networks and their Applications to Chaotic Systems

Abstract

Within the field of machine learning, Echo State Networks (ESN) are a type of neural networks, in which input signals are mapped into higher dimensional spaces, which are connected to outputs. ESNs are used in particular for sequential data, with applications to forecasting. The benefit of using an Echo State Network to predict data is that it is model-free, meaning that there does not need to be any prior knowledge about the data. This is in contrast to using a model-based approach, which requires understanding of the system. This paper will examine the application of ESNs to chaotic systems.