Capacitive neural network with neuro-transistors

TLDR

Resistive neural networks have recently dominated neuromorphic hardware, while capacitive neural networks offer a lower‑power alternative that better emulates neural functions. The authors aim to create neuro‑transistors that emulate neuron soma and axon by integrating dynamic pseudo‑memcapacitors as transistor gates, yielding leaky integrate‑and‑fire dynamics with output signal gain. They achieve this by incorporating dynamic pseudo‑memcapacitors into transistor gates to produce electronic analogs of soma and axon with leaky integrate‑and‑fire behavior and signal amplification. The network demonstrates Hebbian‑like associative learning via non‑volatile pseudo‑memcapacitive synapses and successfully classifies input signals in a fully integrated capacitive neural network.

Abstract

Experimental demonstration of resistive neural networks has been the recent focus of hardware implementation of neuromorphic computing. Capacitive neural networks, which call for novel building blocks, provide an alternative physical embodiment of neural networks featuring a lower static power and a better emulation of neural functionalities. Here, we develop neuro-transistors by integrating dynamic pseudo-memcapacitors as the gates of transistors to produce electronic analogs of the soma and axon of a neuron, with "leaky integrate-and-fire" dynamics augmented by a signal gain on the output. Paired with non-volatile pseudo-memcapacitive synapses, a Hebbian-like learning mechanism is implemented in a capacitive switching network, leading to the observed associative learning. A prototypical fully integrated capacitive neural network is built and used to classify inputs of signals.