Differentiable temporal point processes for spiking neural networks

Abstract

A method performs a Differentiable Point Process (DPP). Generate a first sample sby sampling from a Poisson process with reference to an upper boundof a conditional intensity function representing the DPP given a first set of samples S. Determine whether s>T, output a second set of samplesand train a probabilistic model usingwhen s>T, and perform the next four steps (generate a second sample, add, add, update) and return to the first step (generate a first sample) when s≤T, where T denotes an observation length. Generate a second sample by sampling from a concrete distribution with reference to a parameter of the distribution defined by the conditional intensity function and a temperature τ, given a second set of samples. Add a pair of sand ptoand discard r. Add sto S. Update k to k+1.