NCLIPS-a platform for implementing hybrid expert systems

摘要

The authors categorize connectionist hybrid expert system applications and introduce a hybrid expert system shell which is developed by integrating neural functions into CLIPS. This shell supports hybrid inference chains and neural net training strategies, providing a coherent system for exploiting the advantages of the conventional rule-based expert systems and the neural network techniques in real-life applications. An example application is described to show how neural network applications are made easier under a control strategy programmed in rules.