A Probability Method to Estimate the State of a Digital Resonate-And-Fire Neuron without Running a Simulation

Digital and all-digital resonate-and-fire (RAF) neurons are the newest models for researching spiking neurons and spiking neuron networks (SNN). They are usually researched and developed using time-driven simulation. This type of simulation usually requires powerful computer systems, especially in training networks. Moreover, the digital and all-digital RAF neurons have two cases of firing, leading to hard training for these kinds of neurons compared to leaky integrate-and-fire (LIF) neurons. This research suggested a probability method to find out the firing probability without running a time-driven simulation. By doing so, the research proposed a method to calculate the eigenperiod and pulse width of these types of RAF neurons for machine learning applications.