Forecasting Using Reservoir Computing: The Role of Generalized Synchronization
This work provides a method for researchers and practitioners to more effectively select hyperparameters for Reservoir Computers, potentially improving their performance in time series forecasting.
This paper addresses the challenge of hyperparameter selection in Reservoir Computers (RCs) for time series forecasting. It introduces a pre-training test based on generalized synchronization (GS) to guide hyperparameter selection and proposes using the reproduction of the input system's Lyapunov exponents as a metric for a well-trained RC.
Reservoir computers (RC) are a form of recurrent neural network (RNN) used for forecasting time series data. As with all RNNs, selecting the hyperparameters presents a challenge when training on new inputs. We present a method based on generalized synchronization (GS) that gives direction in designing and evaluating the architecture and hyperparameters of a RC. The 'auxiliary method' for detecting GS provides a pre-training test that guides hyperparameter selection. Furthermore, we provide a metric for a "well trained" RC using the reproduction of the input system's Lyapunov exponents.