Key points are not available for this paper at this time.
Spiking neuron technology holds immense potential for Spiking Neuron Network (SNN) applications, by emulating remarkable computational capabilities of the human brain in the realm of neuromorphic computing. The proposed novel Full Swing Re-configurable Spiking Neuron (FSR-SN) design results in a railto-rail spike which leads to more accurate encoding of information and the re-configurable spike increases computational efficiency of SNNs by reducing quantization errors. This FSR-SN neuron design is employed using the Integrate & Fire (I&F) spiking model, utilizing Metal Oxide Semiconductor (MOS) transistors. This innovative symmetrical design delivers unparalleled energy efficiency, with an ultra-low energy consumption of only 0.482 femtojoules (fJ) per spike, while achieving an impressive spiking frequency of 1.54 GHz. Comprehensive benchmarking against numerous state-of-the-art designs in terms of spiking frequency, energy usage, voltage, and other key metrics has demonstrated the FSR-SN's superior performance, often surpassing the capabilities of existing solutions. This makes the FSR-SN attractive for developing energy-efficient neuromorphic systems and artificial intelligence applications requiring detection of high frequency oscillations.
Kawatkar et al. (Mon,) studied this question.