Next Generation Reservoir Computing

TL;DR

This paper introduces a new approach to reservoir computing using nonlinear vector autoregression, which outperforms traditional methods in efficiency and interpretability for processing dynamical systems.

Contribution

It demonstrates that nonlinear vector autoregression can replace random matrices in reservoir computing, reducing complexity and enhancing performance on benchmark tasks.

Findings

Nonlinear vector autoregression outperforms traditional reservoir computing.

Requires less training data and training time.

Provides more interpretable results.

Abstract

Reservoir computing is a best-in-class machine learning algorithm for processing information generated by dynamical systems using observed time-series data. Importantly, it requires very small training data sets, uses linear optimization, and thus requires minimal computing resources. However, the algorithm uses randomly sampled matrices to define the underlying recurrent neural network and has a multitude of metaparameters that must be optimized. Recent results demonstrate the equivalence of reservoir computing to nonlinear vector autoregression, which requires no random matrices, fewer metaparameters, and provides interpretable results. Here, we demonstrate that nonlinear vector autoregression excels at reservoir computing benchmark tasks and requires even shorter training data sets and training time, heralding the next generation of reservoir computing.