PySpice-Simulated In Situ Learning with Memristor Emulation for Single-Layer Spiking Neural Networks

My research paper, „PySpice-Simulated In Situ Learning with Memristor Emulation for Single-Layer Spiking Neural Networks, “ was published today in the MDPI Electronics journal (IF 2.6), Special Issue: Recent Advances and Related Technologies in Neuromorphic Computing.

Abstract: This paper presents a novel approach to in situ memristive learning by training spiking neural networks (SNNs) entirely within the circuit using memristor emulators in SPICE. The circuit models neurons using Lapicque neurons and employs pulse-based spike encoding to simulate spike-timing-dependent plasticity (STDP), a key learning mechanism in SNNs. The Lapicque neuron model operates according to the Leaky Integrate-and-Fire (LIF) model, which is used in this study to model spiking behavior in memristor-based SNNs. More exactly, the first memristor emulator in PySpice, a Python library for circuit simulation, was developed and integrated into a memristive circuit capable of in situ learning, named the “In Situ Memristive Learning Method for Pattern Classification.” This novel technique enables time-based computation, where neurons accumulate incoming spikes and fire once a threshold is reached, mimicking biological neuron behavior. The proposed method was rigorously tested on three diverse datasets: XPUE, a custom non-dominating 3 × 3 image dataset; a 3 × 5 digit dataset ranging from 0 to 5; and a resized 10 × 10 version of the Modified National Institute of Standards and Technology (MNIST) dataset. The neuromorphic circuit achieved successful pattern learning across all three datasets, outperforming comparable results from other in situ training simulations on SPICE. The learning process harnesses the cumulative effect of memristors, enabling the network to learn a representative pattern for each label efficiently. This advancement opens new avenues for neuromorphic computing and paves the way for developing autonomous, adaptable pattern classification neuromorphic circuits.

The entire code implementation regarding this paper can be found on my GitHub.

Thanks to the Open Science movement and the power of money (which pays the APC fee for open-access journals so that beautiful people like you can read research papers for free), this is my third research paper published this year in the field of neuromorphic computing (after this one and this one).

Extra thoughts: As this is my last research paper published in 2024, I would like to mention that it would be great to research and write a paper about using Physics-Informed Neural Networks (PINNs) to model a printed memristor, as this has not been done yet in the literature (I mentioned this also here a few months ago), with researchers worldwide recently using PINNs to model only non-printed memristors (see here and here), or modeling printed memristors without using PINNs (see here, here, and here). Good news: I started writing a research paper on this topic (only about 30% done), but I don’t know if/when I will finish it, as my new position as a Scientific Officer at ERCEA in Brussels, Belgium is fun and will keep me busy at least for the next five months. Therefore, if you are a PhD student researching the neuromorphic computing domain, I encourage you to use PINNs and model a printed memristor. This idea can also be exciting for bachelor’s and master’s students who plan to write their theses about neuromorphic 3D circuits using (printed) memristors, moving us closer to being able to „print the brain“.