Multi-time scale and high performance in-material reservoir computing using graphene-based ion-gating reservoir

TL;DR

This paper introduces a graphene-based ion-gating reservoir computing system that operates across multiple time scales with high performance, enabling efficient, low-power neuromorphic computing suitable for edge devices.

Contribution

The development of an ion-gel/graphene electric double layer transistor-based reservoir that achieves multi-time scale operation and high accuracy, surpassing traditional physical reservoir computing limitations.

Findings

Operates from 1 MHz to 20 Hz across wide frequency range

Achieves deep learning-level accuracy in chaotic time series prediction

Reduces computational resources to 1/100 of deep learning methods

Abstract

The rising energy demands of conventional AI systems underscore the need for efficient computing technologies like brain-inspired computing. Physical reservoir computing (PRC), leveraging the nonlinear dynamics of physical systems for information processing, has emerged as a promising approach for neuromorphic computing. However, current PRC systems are constrained by narrow operating timescales and limited performance. To address these challenges, an ion-gel/graphene electric double layer transistor-based ion-gating reservoir (IGR) was developed, offering adaptability across multi-time scales with an exceptionally wide operating range from 1 MHz to 20 Hz and high information processing capacity. The IGR achieved deep learning (DL)-level accuracy in chaotic time series prediction tasks while reducing computational resource requirements to 1/100 of those needed by DL. Principal component analysis reveals the IGR's superior performance stems from its high-dimensionality, driven by the ambipolar behavior of graphene and multiple relaxation processes. The proposed IGR represents a significant step forward in providing low-power, high-performance computing solutions, particularly for resource-constrained edge environments.