The many actual and potential applications of expert systems to financial management have received extensive coverage in the financial press. But expert systems, while they can offer help to the financial decision-maker, are poorly suited for many decision environments. In particular, when the decision process is unstructured and reliant on incomplete, confused or error-prone information, expert systems cannot provide the elements of intuition and judgment necessary for solution.
Fortunately, there exists an important alternative artificial-intelligence tool—namely, the artificial neural system (ANS). An ANS is a computer program that simulates the processes by which human learning and intuition take place. Unlike an expert system, an ANS does not rely on a preprogrammed knowledge base. Rather, it learns through experience, and is able to continue learning as the problem environment changes.
The ANS is well suited to deal with unstructured problems, inconsistent information and real-time output. In particular, artificial neural systems are most effectively applied to three tasks—classification, associative memory and clustering. In the area of finance, some potential applications include assessment of bankruptcy risk, identification of arbitrage opportunities and technical and fundamental analysis.
