Iteration over event space in time-to-first-spike spiking neural networks for Twitter (X) bot classification

This study proposes a variant of a time-coding time-to-first-spike spiking neural network (SNN) model with its neurons capable of generating spike trains in response to observed event sequences. This extends an existing model that is limited to generating and observing at most one event per synapse. We explain spike propagation through a model with multiple input and output spikes at each neuron, as well as design training rules for end-to-end backpropagation for event sequence data. The model is trained and evaluated on a Twitter (𝕏) bot detection task where the time of events (tweets and retweets) is the primary carrier of information. This task was chosen to evaluate how the proposed SNN deals with spike train data composed of hundreds of events occurring at timescales differing by almost five orders of magnitude. The impact of various preprocessing steps and training hyperparameter choice on model classification accuracy is analyzed in an ablation study.