Spiking neural networks, an introduction

Abstract

Biological neurons use short and sudden increases in voltage to send information. These signals are more commonly known as action potentials, spikes or pulses. Recent neurological research has shown that neurons encode information in the timing of single spikes, and not only just in their average firing frequency. This paper gives an introduction to spiking neural networks, some biological background, and will present two models of spiking neurons that employ pulse coding. Networks of spiking neurons are more powerful than their non-spiking predecessors as they can encode temporal information in their signals, but therefore do also need different and biologically more plausible rules for synaptic plasticity. You constantly receive sensory input from your environment. You process this information, recognizing food or danger , and take appropriate actions. Not only you; anything that interacts with its environment needs to do so. Mimicking such a seemingly simple mechanism in a robot proofs to be insanely difficult. Nature must laugh at our feeble attempts; animals perform this behaviour with apparent ease. The reason for this mind-boggling performance lies in their neural structure or 'brain'. Millions and millions of neurons are interconnected with each other and cooperate to efficiently process incoming signals and decide on actions. A typical neuron sends its signals out to over 10.000 other neu-rons, making it clear to even to inexpert reader that the signal flow is rather complicated. To put it mildly: we do not understand the brain that well yet. In fact, we do not even completely understand the functioning of a single neuron. The chemical activity of the synapse already proves to be infinitely more complex than firstly assumed. However, the rough concept of how neurons work is understood: neurons send out short pulses of electrical energy as signals, if they have received enough of these themselves. This basically simple mechanism has been moulded into a mathematical model for computer use. Artificial as these computerised neurons are, we refer to them as networks of artificial neurons, or artificial neural networks. We will sketch a short history of these now; the biological background of the real neuron will be drawn in the next chapter. Generations of artificial neurons Artificial neural networks are already becoming a fairly old technique within computer science; the first ideas and models are over fifty years old. The first generation of artificial neural networks consisted of McCulloch-Pitts threshold neu-rons [15], a conceptually very simple …

Extracted Key Phrases

3 Figures and Tables

Showing 1-10 of 26 extracted citations