An important bottleneck in present-day neuromorphic hardware is its reliance on synaptic addition, which limits the achievable degree of parallelization and thus processing throughput. We present a network of monostable multivibrator timers, whose synaptic inputs are simply OR-ed together, thus mitigating the synaptic addition bottleneck. Monostable multivibrators are simple timers which are easily implemented using counters in digital hardware and can be interpreted as non biologically-inspired spiking neurons. We show how fully binarized event-driven recurrent networks of monostable multivibrators can be trained to solve classification tasks. Our training algorithm resolves temporally overlapping input events. We demonstrate our approach on the MNIST handwritten digits, Google Soli radar gestures, IBM DVS128 gestures and Yin-Yang classification tasks. The estimated energy consumption for the MNIST handwritten digits task, excluding the final linear readout layer, is 855pJ per inference for a test accuracy of 98.61\% for a reconfigurable network of 500 units, when mapped to the TSMC HPC+ 28nm process.
Keuninckx, L, Hartmann, M, Detterer, P, Safa, A, Mommen, W & Ocket, I 2026, 'On training networks of monostable multivibrator timer neurons', Neural Networks, vol. 194, 108092. https://doi.org/10.1016/j.neunet.2025.108092
Keuninckx, L., Hartmann, M., Detterer, P., Safa, A., Mommen, W., & Ocket, I. (2026). On training networks of monostable multivibrator timer neurons. Neural Networks, 194, Article 108092. https://doi.org/10.1016/j.neunet.2025.108092
@article{ba4413a659e940d48c57a55e7f246ba9,
title = "On training networks of monostable multivibrator timer neurons",
abstract = "An important bottleneck in present-day neuromorphic hardware is its reliance on synaptic addition, which limits the achievable degree of parallelization and thus processing throughput. We present a network of monostable multivibrator timers, whose synaptic inputs are simply OR-ed together, thus mitigating the synaptic addition bottleneck. Monostable multivibrators are simple timers which are easily implemented using counters in digital hardware and can be interpreted as non biologically-inspired spiking neurons. We show how fully binarized event-driven recurrent networks of monostable multivibrators can be trained to solve classification tasks. Our training algorithm resolves temporally overlapping input events. We demonstrate our approach on the MNIST handwritten digits, Google Soli radar gestures, IBM DVS128 gestures and Yin-Yang classification tasks. The estimated energy consumption for the MNIST handwritten digits task, excluding the final linear readout layer, is 855pJ per inference for a test accuracy of 98.61\% for a reconfigurable network of 500 units, when mapped to the TSMC HPC+ 28nm process.",
author = "Lars Keuninckx and Matthias Hartmann and Paul Detterer and Ali Safa and Wout Mommen and Ilja Ocket",
note = "Publisher Copyright: {\textcopyright} 2025 Elsevier Ltd",
year = "2026",
month = feb,
doi = "10.1016/j.neunet.2025.108092",
language = "English",
volume = "194",
journal = "Neural Networks",
issn = "0893-6080",
publisher = "Elsevier Limited",
}