DocumentCode :
3610603
Title :
To Spike or Not to Spike: That Is the Question
Author :
Maass, Wolfgang
Author_Institution :
Inst. for Theor. Comput. Sci., Graz Univ. of Technol., Graz, Austria
Volume :
103
Issue :
12
fYear :
2015
Firstpage :
2219
Lastpage :
2224
Abstract :
Both the brain and digital computers process information, but they do this in completely different ways. Neurons in the brain transmit information not through bits, but through spikes. Spikes are short voltage increases that are generated near the cell body of a neuron, with average spike rates below 10 Hz. These spikes are transmitted via fine axonal fibers and synapses to about 10 000 other neurons. Neurons also differ in another fundamental aspect from processors in a digital computer: they produce spikes according to stochastic rather than deterministic rules. This article discusses recent progress in understanding how complex computations can be carried out with such stochastically spiking neurons. Other recent developments suggest that spike-based neural networks can be emulated by neuromorphic hardware at a fraction of the energy consumed by current digital computing hardware. Can both developments be merged to provide a blueprint for substantially more energy-efficient computing devices? Explores these issues and examines the viability of such a merger.
Keywords :
bioinformatics; brain; digital computers; neural nets; axonal fiber; brain; digital computer; neuromorphic hardware; neurons; spike-based neural network; Biological neural networks; Brain models; Digital computers; Energy efficiency; Neural networks; Neurons; Optical fiber devices; Optical fiber networks; Program processors; Voltage measurement;
fLanguage :
English
Journal_Title :
Proceedings of the IEEE
Publisher :
ieee
ISSN :
0018-9219
Type :
jour
DOI :
10.1109/JPROC.2015.2496679
Filename :
7329679
Link To Document :
بازگشت