Document Preview
  • Full Text
  • Scholarly Journal

Some potential applications of artificial neural systems in financial management

Full text preview


An artificial neural system (ANS) is an artificial-intelligence tool that attempts to simulate the physical process upon which intuition is based--that is, by simulating the process of adaptive biological learning. Theoretically, an ANS is capable of producing a proper response to a given problem (or the best possible response, when more than one response is applicable) even when the information is fuzzy or incomplete, or when no predefined procedure for solving the problem is available. This paper describes some areas of financial management where the artificial neural systems could be very useful.


An Artificial Neural System (ANS) models, in a very simplified way, the biological systems of the human brain. It does so by grouping the computer units that function like the major component of the human brain--the neurons. Simulated neurons serve as the basic functional unit of the ANS in much the same way that the binary electronic switches serve as the basic unit in digital computers.

Modern information technologies make the development of artificial neural systems a step closer to reality. The ANSs are particularly efficient in doing such operations as image recognition, forecasting, text retrieval, and optimization. Most of these are difficult to do with conventional single-processor solutions . A few financial applications that involve such operations as forecasting and optimization have been shown as potential candidates for adapting to this new technology.


Like the brain itself, an ANS depends for power and speed on the simultaneous functioning of its individual neural unit. Neural net systems differ from traditional computer applications in many ways. Neural nets are not programmed in the traditional sense. An ANS is not provided with quantitative descriptions of objects or patterns to be recognized, or with logical criteria for distinguishing such objects from similar objects. Instead, it is presented...