Artificial Intelligence

Abstract

Artificial intelligence is an engineering method that simulates the structures and operating principles of the human brain. Much is unknown of how the brain trains itself to process information, but we do know that brain cells, called neurons, can be activated to send an electric signal through long thin stands called axons. At the end of the axon a structure called the synapse connects the axon with a connected neuron, and provides it with excitatory/inhibitory imput or when the signal is too weak no imput at all. Learning processes in the brain is thought to take place by repeated similar electric signals at similar places giving rise to similar outcomes observed by the brain. This principle can be modeled by artificial neural networks software using observed variables as artificial signals. Software is available in SPSS, MATLAB and so forth: in the current paper SPSS version 17.0, with neural network add-on has been applied (WWW.SPSS.COM). Artificial neural networks are different from traditional statistics that usually assumes Gaussian curve distributions for making predictions from the data. In practice data, sometimes, do not follow Gaussian distributions, and, for that purpose, distribution-free methods, like non-parametric tests and Monte Carlo methods, have been developed. The artificial neural network is another distribution-free method based on layers of artificial neurons that transduce imputed information. It has been recognized to have a number of advantages including the possibility to process imperfect data, and complex non linear data (Stergiou and Siganos 2004). The current chapter reviews the principles, procedures, and limitations of BP artificial neural networks for a non-mathematical readership.