Applications of spiking neural networks
暂无分享,去创建一个
We are pleased to introduce this issue of Information Processing Letters pre-senting state-of-the-art articles on Applications of Spiking Neural Networks.Spiking neural networks are a class of neural networks that is increasinglyreceiving attention as both a computationally powerful and biologically moreplausible model of distributed computation. Much work so far has focusedon fundamental issues like computational complexity, biologically plausiblemodels, effects of biological learning rules and such.In this issue, there are five articles that consider how spiking neural networkscan be used towards applications. We will discuss them in turn.One intriguing proposal for how to make use of networks of spiking neuronshas been the Liquid State Machine, by Maass, Natschl¨ager & Markram [3].Maass et al. realized that a randomly connected network of spiking neuronseffectively implements a complex temporal filter through the intricacies ofreverberating activity and synaptic dynamics. Given a temporally extendedinput, like speech, the collective activity of the network can be described as atrajectory through a high-dimensional state space, and this trajectory shouldbe identifiably specific for the input at hand. A simple “read-out” decodershould then be sufficient to classify the temporal pattern.In the paper “Isolated Word Recognition With The Liquid State Machine:A Case Study”, Verstraeten, Schrauwen, Stroobandt and Van Campenhoutstudy speech recognition in the Liquid State Machine (LSM). For a standardvocabulary set, they test a variety of temporal encodings that are fed into anLSM, and a subsequent simple linear decoder classifies each particular pattern.Surprisingly, Verstraeten et al. only find (encouragingly good) performance forthe LSM when the encoding back-end approximates the encoding scheme ofthe inner ear.Learning rules equivalent to those employed in traditional sigmoidal neuralnetworks have also been derived for (layered) spiking neural networks, for
[1] Henry Markram,et al. Real-Time Computing Without Stable States: A New Framework for Neural Computation Based on Perturbations , 2002, Neural Computation.
[2] Sander M. Bohte,et al. Error-backpropagation in temporally encoded networks of spiking neurons , 2000, Neurocomputing.
[3] Peter Dayan,et al. Probabilistic Computation in Spiking Populations , 2004, NIPS.
[4] Rajesh P. N. Rao. Hierarchical Bayesian Inference in Networks of Spiking Neurons , 2004, NIPS.