An object-oriented hybrid environment for integrating neural networks and experts systems

With the emerging realization that most complex real-world problems are difficult to solve by either symbolic or adaptive paradigms, there is great interest in combining the strengths of individual techniques (such as neural networks and expert systems), from these opposing forms of information processing. The object-oriented hybrid environment described allows the strengths of these contending processing paradigms to be combined for solving complex problems. The use of object-oriented methods brings with it many attributes and advantages for constructing hybrid systems. This approach allows each paradigm to be represented as an object, which can communicate with other paradigms via a message passing mechanism. The operation details of the neural-symbolic environment are illustrated and tested, with an application from the financial arena of profit trend analysis.<<ETX>>