Striking a Chord: Advancing Timeseries Data Recognition with Spiking Neural Networks

In recent years, Spiking Neural Networks (SNNs) have emerged as a promising direction in neuromorphic computing, providing a more biologically plausible framework than their conventional artificial neural network counterparts. This presentation argues the unique inherent capabilities to process and recognize temporal patterns make SNNs an excellent candidate for time series analysis in physics.

I explore the computational advantages of SNN by showcasing their application on temporal physical phenomena. In particular we tackle auditory recognition by delving into their capacity to discern intricate auditory patterns, such as the classification of musical chords. Using the neuromorphic framework developed by TENNLabs from the University of Tennessee, we were able to successfully create a SNN which categories piano chords into one of the 12 western tonal notes with varying degrees of success.