Perspectives of Artificial Neural Network

Abstract

A majority of information processing today is carried out by digital computers. Recent Neuro psychological experiments have shed considerable light on the structure of brain, and even in fields, such as cognitive science, which study human information processing process at the macro level. Research’s in the field of mathematical science and physics is also concentrating more on the mathematical analysis of systems comprising multiple elements that interact in complex ways. These factors gave birth to a major research trend aimed at clarifying the structures and operating principles inherent in the information processing device based on these structures and operating principles. The term neuro-computing is used to refer to the information engineering aspects of this research. ANN is superior for pattern recognition and is able to deal with any model whereas statistical methods require randomness. The old adage of garbage in, garbage out holds especially true for ANN modelling. A well known case in which an ANN learned the incorrect model involved the identification of a person’s sex from a picture of his/her face. The ANN application was trained to identify a person as either male or female by being shown various pictures of different persons’ faces. At first, researchers thought that the ANN had learnt to differentiate the face of a male from that of a female, by identifying the visual features, of a person’s face. However it was later discovered that the pictures used as input data showed all the male persons’ heads nearer to the edge of the top end of the pictures, presumably due to a bias of taller males in the data than females. The ANN model had therefore learned to differentiate the sex of a person by the distance his/her head is from the top edge of a picture rather than by identifying his/her visual facial features.