Reducing hyperparameter dependence by external timescale tailoring

TL;DR

This paper demonstrates that tailoring reservoir timescales to specific tasks reduces the need for hyperparameter tuning in reservoir computing, especially for temporal tasks requiring memory of past inputs.

Contribution

It introduces methods for incorporating task-specific timescales into reservoir computing, decreasing hyperparameter sensitivity and improving performance.

Findings

Hyperparameter tuning can be minimized with tailored timescales.

Task-specific timescales improve reservoir performance on temporal tasks.

Methods are effective for both time-multiplexed and spatially multiplexed reservoirs.

Abstract

Task specific hyperparameter tuning in reservoir computing is an open issue, and is of particular relevance for hardware implemented reservoirs. We investigate the influence of directly including externally controllable task specific timescales on the performance and hyperparameter sensitivity of reservoir computing approaches. We show that the need for hyperparameter optimisation can be reduced if timescales of the reservoir are tailored to the specific task. Our results are mainly relevant for temporal tasks requiring memory of past inputs, for example chaotic timeseries prediciton. We consider various methods of including task specific timescales in the reservoir computing approach and demonstrate the universality of our message by looking at both time-multiplexed and spatially multiplexed reservoir computing.