This paper presents a new supervised learning algorithm (Spike‐Comp) with an adaptive compact structure for Spiking Neural Networks (SNNs). SpikeComp consists of two layers of spiking neurons: an encoding layer which temporally encodes real valued features into spatio-temporal spike patterns, and an output layer of dynamically grown neurons which perform spatio-temporal pattern classification. The weights between the neurons in the encoding layer and the new added neuron in the output layer are initialised based on the precise spiking times in the encoding layer. New strategies are proposed to either add a new neuron, or update the network parameters when a new sample is presented to the network. The proposed learning algorithm was demonstrated on several benchmark classification datasets and the obtained results show that Spike Comp can perform pattern classification with a comparable performance and a much compact network structure compared with other existing SNN training algorithm.
CITATION STYLE
Wang, J., Belatreche, A., Maguire, L. P., & McGinnity, T. M. (2015). Spikecomp: An evolving spiking neural network with adaptive compact structure for pattern classification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9490, pp. 259–267). Springer Verlag. https://doi.org/10.1007/978-3-319-26535-3_30
Mendeley helps you to discover research relevant for your work.