Reconfigurable cascaded thermal neuristors for neuromorphic computing

TL;DR

This paper introduces reconfigurable thermal neuristors based on vanadium dioxide, demonstrating their potential for scalable, energy-efficient neuromorphic computing through thermal interactions that mimic biological neural dynamics.

Contribution

It presents a novel class of thermal neuristors utilizing insulator-metal transition, enabling reconfigurable dynamics and cascaded information flow solely via thermal processes.

Findings

Demonstrated inhibitory functionality in a single oxide device

Achieved cascaded thermal interactions for information flow

Developed a theoretical model matching experimental results

Abstract

While the complementary metal-oxide semiconductor (CMOS) technology is the mainstream for the hardware implementation of neural networks, we explore an alternative route based on a new class of spiking oscillators we call thermal neuristors, which operate and interact solely via thermal processes. Utilizing the insulator-to-metal transition in vanadium dioxide, we demonstrate a wide variety of reconfigurable electrical dynamics mirroring biological neurons. Notably, inhibitory functionality is achieved just in a single oxide device, and cascaded information flow is realized exclusively through thermal interactions. To elucidate the underlying mechanisms of the neuristors, a detailed theoretical model is developed, which accurately reflects the experimental results. This study establishes the foundation for scalable and energy-efficient thermal neural networks, fostering progress in brain-inspired computing.