A Functional Spiking Neural Network of Ultra Compact Neurons

Abstract

We demonstrate that recently introduced ultra-compact neurons (UCN) with a minimal number of components can be interconnected to implement a functional spiking neural network. For concreteness we focus on the Jeffress model, which is a classic neuro-computational model proposed in the 40’s to explain the sound directionality detection by animals and humans. In addition, we introduce a long-axon neuron, whose architecture is inspired by the Hodgkin-Huxley axon delay-line and where the UCNs implement the nodes of Ranvier. We then interconnect two of those neurons to an output layer of UCNs, which detect coincidences between spikes propagating down the long-axons. This functional spiking neural neuron circuit with biological relevance is built from identical UCN blocks, which are simple enough to be made with off-the-shelf electronic components. Our work realizes a new, accessible and affordable physical model platform, where neuroscientists can construct arbitrary mid-size spiking neuronal networks in a lego-block like fashion that work in continuous time. This should enable them to address in a novel experimental manner fundamental questions about the nature of the neural code and to test predictions from mathematical models and algorithms of basic neurobiology research. The present work aims at opening a new experimental field of basic research in Spiking Neural Networks to a potentially large community, which is at the crossroads of neurobiology, dynamical systems, theoretical neuroscience, condensed matter physics, neuromorphic engineering, artificial intelligence, and complex systems.

Quantum materials for energy-efficient neuromorphic computing

“…The thyristor,a conventional semiconductor electronics component, 159 which was recently recognized to have memristive properties that are qualitative similar to Mott materials, demonstrated coupling with leaky-integrate-and-fire artificial neurons. 160,161 Another way to implement electric coupling is capacitively. The simplest spiking neurons based on Mott materials are spiking oscillators, where a Mott device is connected in parallel to a capacitor.…”

Implementation of a Minimal Recurrent Spiking Neural Network in a Solid-State Device

“…The output of the UCN circuit implements an "axon," which strengthens the action-potential spike on R 3 . This is done so that the UCN spikes may drive other neurons that can be directly connected downstream, allowing networks to be built like construction with Legolike blocks [12,28]. The axon can be simply implemented by means of a transistor Q 1 (and a +5 V source).…”

“…1(c)] that consists of a chain of identical UCNs, as described in Ref. [28]. The long-axon configuration is a transmission line, much in the spirit of the Hodgkin-Huxley model for the propagation of the action potential along the squid giant axon [29].…”

“…It permits a delay to be implemented in the form of nδt, where δt is the delay of the signal introduced by each block of the long axon, and n is the position of the block in the chain (δt is essentially the discharge time discussed above, also see Ref. [28]). Thus, to implement a delay of t = nδt, it is sufficient to take the spike signal from the nth Ranvier node of the axon chain [for instance, t = 4δt in Fig.…”

“…Thus, to implement a delay of t = nδt, it is sufficient to take the spike signal from the nth Ranvier node of the axon chain [for instance, t = 4δt in Fig. 1(c)] [28]. Here, however, we do not implement the long axon but, for convenience, we implement the delay with a standard electronic-dataacquisition instrument, as we describe in the Supplemental Material [14,30].…”