Abstract: An expert system and processing method employing a three level hierarchical knowledge base that has a plurality of nodes coupled together. An uppermost level comprises a behavioral knowledge level, a middle level comprises a structural knowledge level and a lowermost level is an action level. Inference processing proceeds from the behavioral knowledge level through the structural knowledge level to a leaf node of the action level. One or more non-inferentially assessable utilities may be associated with the inference kernel of the expert system for accessing during inferential processing within the hierarchically structured knowledge base. A knowledge editor implements guidelines that structure received information in the desired hierarchical three level configuration. Further, interactive multimedia/hypermedia systems and methods are presented, which might be expert system based. Training and certification applications of the multimedia/hypermedia system/method are given.

Abstract: A real-time learning (RTL) neural network is capable of indicating when an input feature vector is novel with respect to feature vectors contained within its training data set, and is capable of learning to generate a correct response to a new data vector while maintaining correct responses to previously learned data vectors without requiring that the neural network be retrained on the previously learned data. The neural network has a sensor for inputting a feature vector, a first layer and a second layer. The feature vector is supplied to the first layer which may have one or more declared and unused nodes. During training, the input feature vector is clustered to a declared node only if it lies within a hypervolume defined by the declared node's automatically selectable reject radius, else the input feature vector is clustered to an unused node. Clustering in overlapping hypervolumes is determined by a decision surface.

Abstract: In this network (R) all the cells have identical construction. Each cell is assigned a code word which defines its function which it is to provide for within the network. With a view to the configuring of the cells (Cx,y), means (1) iteratively cause the translation into each of the cells of a genome formed of a train of the successive code words.The network comprises at least one individualized functional organ (O1, O2, O3, . . . ) composed of a plurality of cells which are all characterized by a first predetermined code (<MHC>) forming part of said code word and formed by a number identifying the organ (O1, O2, O3, . . . ) so that these cells can collectively provide for the functionality of this organ. The cells of the latter are distributed as internal cells defined by a first predetermined value of a status code also forming part of the code word, and as membrane cells defined by a second predetermined value of the second code.

Abstract: In the design and implementation of neural networks, training is determined by a series of architectural and parametric decisions. A method is disclosed that, using genetic algorithms, improves the training characteristics of a neural network. The method begins with a population and iteratively modifies one or more parameters in each generation based on the network with the best training response in the previous generation.