Class scheduling by neurocomputing: a comparison with the expert system approach

A university class scheduling support system based on neurocomputing is described. The structure and effectiveness of this system are described. The scheduling problem is divided into a number of steps, and a Hopfield-type network is applied to each step. The results confirm that a good system can be constructed. The neurocomputing approach is compared to the rule-based expert system approach previously applied to the same problem. It is concluded that each has advantages and that, in building design-type expert systems, a trend will be to combine the neurocomputing and rule-based approaches.<
>