Emulating short-term synaptic dynamics with memristive devices

Eleni Vasilaki,Giacomo Indiveri,Radu Berdan,Ali Khiat,Themistoklis Prodromakis,Alexandru Serb

doi:10.1038/srep18639

Abstract

Neuromorphic architectures offer great promise for achieving computation capacities beyond conventional Von Neumann machines. The essential elements for achieving this vision are highly scalable synaptic mimics that do not undermine biological fidelity. Here we demonstrate that single solid-state TiO2 memristors can exhibit non-associative plasticity phenomena observed in biological synapses, supported by their metastable memory state transition properties. We show that, contrary to conventional uses of solid-state memory, the existence of rate-limiting volatility is a key feature for capturing short-term synaptic dynamics. We also show how the temporal dynamics of our prototypes can be exploited to implement spatio-temporal computation, demonstrating the memristors full potential for building biophysically realistic neural processing systems.

Highlights

In this work we presented detailed and quantitative parallels between memristive devices and biophysically realistic models of synaptic dynamics
In particular we showed how meta-stable memory transitions that are typically seen as non-ideal effects that contribute to the large variability observed in emerging ReRAM, are a salient feature for the establishment of truly biomimetic synapses that can faithfully reproduce short-term synaptic dynamics
In addition to reproducing plasticity mechanisms at a phenomenological level, these devices can be biased to exhibit both stochastic properties and biologically plausible temporal dynamics. These features can be exploited for developing non-von Neuman computing architectures in which memory and computation are co-localized, and where massively parallel circuits can process signals in real-time. Their ability to implement biologically realistic time constants would enable the construction of neural computing systems that can efficiently process real-world biologically relevant sensory signals and interact with the environment

Summary

Introduction

In this work we presented detailed and quantitative parallels between memristive devices and biophysically realistic models of synaptic dynamics. In particular we showed how meta-stable memory transitions that are typically seen as non-ideal effects that contribute to the large variability observed in emerging ReRAM, are a salient feature for the establishment of truly biomimetic synapses that can faithfully reproduce short-term synaptic dynamics. In addition to reproducing plasticity mechanisms at a phenomenological level, these devices can be biased to exhibit both stochastic properties and biologically plausible temporal dynamics.

Results

Conclusion