This paper will provide an introductory level discussion of neural networks within the field of artificial intelligence. This discussion will briefly cover the history of the neural network as well as recent advances within this field. In addition, several real world applications of neural networks will be discussed.
The primary goal in the field of artificial intelligence is to construct a machine with an intellect comparable to that of a human. This pursuit of an artificial intelligence has had a long history. Several different approaches have been attempted as a result of this goal. In particular, the study of neural networks has evolved from this pursuit for an intelligent machine.
The field of neural networks involves a new approach to computing that uses mathematical structures with the ability to learn (Zsolutions). These methods were inspired by investigations into modeling nervous system learning (Zsolutions). For example, neurons in the human brain are used to transmit data back and forth to each other. Artificial neural networks use this same technique to process various kinds of information (Fu, p 4).
There are a wide variety of applications in which neural networks can be utilized. Primarily, they should be used in areas where standard techniques fail to give satisfactory results (Zsolutions). Neural networks are applied best in situations where information needs to be determined faster and with more efficiency. In addition, neural networks outperform other artificial intelligence approaches in areas where more detail can be learned from inputted data (Zsolutions).
The technology of neural networks has been in existence for approximately forty years (CIO). By the 1940s, research on the behavior of the human neuron was documented enough for psychologists and mathematicians to make an attempt at a mathematical theory of the neuron (Merlin). This is essentially the way a human neuron works:
Biological neurons transmit electrochemical signals from other neurons through special junctions called synapses. Some inputs tend to excite the neuron; others tend to inhibit it. When the cumulative effect exceeds a threshold, the neuron fires and sends a signal down to other neurons. (Fu, p8)
Shortly after this information was documented, progress towards the development of the first artificial neural network began.
Work published by Warren McCulloch and Walter Pitts in 1943 led to the creation of the first artificial neural network. Their work laid the foundation from which modern day neural networks are based (Merlin). They constructed an artificial neuron network consisting of three kinds of neurons:
1. Receptor Neurons - input neurons that receive the impulse to fire from a sensor.
2. Central Neurons - inner neurons which are synapsed onto from receptor and other neurons and synapse onto output and other neurons.
3. Effector Neurons - receive impulses...