An evolutionary strategy for learning in fuzzy networks
C. Duarte, J.A.B. Tome · 2002
Sine the pioneer work of White the application of artificial neural networks to finance has enjoyed an exponential growth in research and publications. The evidence accumulated over the last decade indicates that the success of the financial application of artificial neural networks depend on its design. Due to the nature of artificial networks it's very hard for the experts in the field of the application to share their knowledge with the system. A rule based approach would make this interchange easier. Lin and Lee introduced a system integrating neural networks and fuzzy logic that displays a low level learning capability and high-level thinking and reasoning ability. This paper presents a modification to the learning methodology proposed by Lin and Lee. A Genetic Algorithm is employed to train the network, both the connections between nodes and the node parameters. The resulting system is applied to the daily forecast of foreign currency.