The use of a modified Hopfield neural network to perform truth maintenance

In order for a computer to be able to be considered even rudimentarily intelligent, it must be able to store knowledge in a sensible and useful manner. One way that knowledge can be stored is in a belief graph, which keeps track of a set of interdependent beliefs. The consistency of a belief graph can be maintained by a reason maintenance system. This thesis discusses a reason maintenance algorithm based on a Hopfield neural net that provides several advantages over past truth maintenance algorithms.