Memristive Silicon Spiking Neurons
We invented a new type of electronic circuit for a spiking neuron with a minimal number of conventional electronic components. We need just three elements to implement a leaky-integrate-and-fire (LIF) spiking neuron: a resistor (leak), a capacitor (integrate) and a thyristor (fire). This unprecedented simplicity was made posible by the insight of realizing that a thyristor+resistor behaves as a memristor , which is an emerging concept relevant to Neuromorphic Engineering. . The photo is the actual circuit that we fabricated with out-of-the-shelf electronic components, showing several artificial electronic neurons.
Recent Articles
An ultra-compact leaky-integrate-and-fire model for building spiking neural networks. M. Rozenberg, O. Schneegans, P. Stoliar; Scientific Reports, Nature Publishing Group, 9, 11123(2019).
Biologically relevant dynamical behaviors realised in an ultra-compact neuron model. P. Stoliar, O. Schneegans and M. Rozenberg; Frontiers in Neurosciences 14, 421 (2020).
A Functional Spiking Neural Network of Ultra Compact Neurons. P. Stoliar, O. Schneegans and M. J. Rozenberg; Frontiers in Neurosciences 15, 102 (2021).
Implementation of a minimal recurrent spiking neural network in a solid state device. P. Stoliar, O. Schneegans and M. J. Rozenberg Phys. Rev. App. 16, 034030 (2021)
A spiking neuron implemented in VLSI, P. Stoliar, I. Akita, O. Schneegans, M. Hioki, and M J Rozenberg; J. Phys. Commun. 6 021001(2022)
Solid State Neuroscience: Spiking Neural Networks as Time Matter. M. Rozenberg Journal of Physics: Conference Series 2533 (1), 012007 (2023)
NEW Bursting dynamics in a spiking neuron with a memristive voltage-gated channel J Wu, K Wang, O Schneegans, P Stoliar, and M Rozenberg; Neuromorphic Computing and Engineering (in press)