A Deep-Learning-Based Method For Spike Sorting With Spiking Neural Network

Abstract

Spike sorting is a technique of detecting and assigning signals generated by the neurons of the brain and classifying which spike belongs to which neurons. Spike sorting is usually done by a clustering algorithm that groups the spikes coming from the same neuron in a pre-defined feature space. Artificial Neural Networks (ANNs), one of the clustering algorithms for spike sorting, are widely used and give a promising performance in many areas. However, ANNs require a large amount of data for training with heavy computational costs because of the architecture of ANN and back-propagation. This paper presents a Spiking Neural Networks (SNNs), the third generation of Artificial Neural Networks, trained with an unsupervised Spike-Timing-Dependent-Plasticity (STDP) learn-ing rule that modifies synaptic strength depending on the relative timing of spikes fired. The objective of this work is to investigate the performance of SNNs on spike sorting. Many automatic spike sorting algorithms such as template matching, K-means clustering and Artificial Neural Networks have been proposed for spike sorting to replace human intervention. Hence, the result of this work and the traditional automatic spike sorting methods will be shown and compared at the end of this paper. The publicly available simulated spike dataset from the University of Leicester which is commonly used in spike signal research is used for this study.