Iteration over event space in time-to-first-spike spiking neural networks for Twitter bot classification
This addresses bot detection on Twitter by handling temporal event data, but it is incremental as it builds on existing SNN frameworks.
The study tackled processing time-varying information in spiking neural networks by extending time-to-first-spike models with multiple spikes per neuron and end-to-end backpropagation training, achieving evaluation on a Twitter bot detection task with events spanning five orders of magnitude timescales.
This study proposes a framework that extends existing time-coding time-to-first-spike spiking neural network (SNN) models to allow processing information changing over time. 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. This strategy enables us to process information changing over time. The model is trained and evaluated on a Twitter 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 parameters on model properties, performance and training-time stability is analyzed.