Work

Spiking flip-flop memory in resonant tunnelling diode neurons

Year: 2024

Type: article

Abstract: We report a spiking flip-flop memory mechanism allowing to controllably switch between neural-like excitable spike-firing and quiescent dynamics in a resonant tunneling diode (RTD) neuron under low-amplitude (<150 mV pulses) and high-speed (ns-rates) inputs pulses. We also show that the timing of the set-reset input pulses is critical to elicit switching responses between spiking and quiescent regimes in the system. The demonstrated flip-flop spiking memory, in which spiking regimes can be controllably excited, stored and inhibited in RTD-neurons, via specific low-amplitude, high speed signals (delivered at proper time instants) offers high promise for RTD-based spiking neural networks, with potentials to extend further to optoelectronic implementations where RTD neurons and RTD memory elements are deployed alongside for fast and efficient photonic-electronic neuromorphic computing and Artificial Intelligence (AI) hardware.

Source: Strathprints: The University of Strathclyde institutional repository (University of Strathclyde)