Reservoir Computing with Diverse Timescales for Prediction of Multiscale Dynamics

Abstract

Machine learning approaches have recently been leveraged as a substitute or an aid for physical/mathematical modeling approaches to dynamical systems. To develop an efficient machine learning method dedicated to modeling and prediction of multiscale dynamics, we propose a reservoir computing (RC) model with diverse timescales by using a recurrent network of heterogeneous leaky integrator (LI) neurons. We evaluate computational performance of the proposed model in two time series prediction tasks related to four chaotic fast-slow dynamical systems. In a one-step-ahead prediction task where input data are provided only from the fast subsystem, we show that the proposed model yields better performance than the standard RC model with identical LI neurons. Our analysis reveals that the timescale required for producing each component of target multiscale dynamics is appropriately and flexibly selected from the reservoir dynamics by model training. In a long-term prediction task, we demonstrate that a closed-loop version of the proposed model can achieve longer-term predictions compared to the counterpart with identical LI neurons depending on the hyperparameter setting.

arXiv comment7 pages, 6 figures, supplemental material (10 pages)

References in corpus (13)

• Recent Advances in Physical Reservoir Computing: A Review
• Optoelectronic Reservoir Computing
• Integrating Machine Learning and Multiscale Modeling: Perspectives, Challenges, and Opportunities in the Biological, Biomedical, and Behavioral Sciences
• Regularization of Synchronized Chaotic Bursts
• Hybrid Forecasting of Chaotic Processes: Using Machine Learning in Conjunction with a Knowledge-Based Model
• Discovery of Nonlinear Multiscale Systems: Sampling Strategies and Embeddings
• Predicting the spatiotemporal diversity of seizure propagation and termination in human focal epilepsy
• An extension of the Lyapunov analysis for the predictability problem
• Performance boost of time-delay reservoir computing by non-resonant clock cycle
• The Emergence of Modularity in Biological Systems
• Exploiting Multiple Timescales in Hierarchical Echo State Networks
• Optimal responsiveness and collective oscillations emerging from the heterogeneity of inhibitory neurons
• Effective models and predictability of chaotic multiscale systems via machine learning

Cited by in corpus (5)

• Effect of temporal resolution on the reproduction of chaotic dynamics via reservoir computing
• A Dynamic Systems Approach to Modelling Human-Machine Rhythm Interaction
• Phase transitions from linear to nonlinear information processing in neural networks
• A novel Reservoir Architecture for Periodic Time Series Prediction
• Predicting unobserved climate time series data at distant areas via spatial correlation using reservoir computing