Nonlinear iontronic signal processing with neuromorphic Spike Rate-Dependent Plasticity

TL;DR

This paper introduces an integrated iontronic memristor circuit that mimics biological Spike Rate-Dependent Plasticity, enabling nonlinear frequency-based classification of auditory signals, advancing neuromorphic computing capabilities.

Contribution

The work presents a novel integrated iontronic memristor circuit with heterogeneous internal timescales that exhibits nonlinear frequency response and biological SRDP, demonstrated for auditory data classification.

Findings

Circuit exhibits nonlinear frequency response similar to biological SRDP

Successfully classifies insect sounds that are linearly inseparable

Experimental results align with theoretical model predictions

Abstract

We present an integrated iontronic memristor circuit that reproduces biologically inspired Spike Rate-Dependent Plasticity (SRDP) and functions as a physical nonlinear frequency kernel, which we demonstrate can be used to classify natural auditory data. The fluidic circuit integrates two parallel memristive membranes containing short and long conical memristive channels with opposite orientations, giving rise to heterogeneous internal timescales and different potentiation responses. As a result, the circuit exhibits a nonlinear frequency response in which low-frequency inputs decrease the overall conductance, whereas higher-frequency inputs increase it, thereby emulating biological SRDP. Our experimental measurements are inspired by and consistent with predictions of a theoretical model. We demonstrate the functionality of the device by separating encoded sound signals from different insects that cannot be linearly separated. By unifying theoretical predictions with experimental realisation of coupled iontronic memristors, this work moves beyond isolated components and demonstrates how heterogeneous iontronic dynamics can unlock nonlinear time-series processing capabilities, essential for future iontronic neuromorphic computing.