Search

The phrase-structure grammar scheme used for specifying string languages is extended to structures called plexes composed of symbols with an arbitrary number of ''attaching points.'' Classes of plex languages that parallel existing classes of string languages are defined, and these classes are shown to be distinct. Context-free grammars for languages of chemical structures, logic diagrams, electrical circuits, and flowcharts are given. Plex languages are used to specify the interconnection of encoded geometric curves to form mesh-like line patterns, and methods are given by which languages of such line patterns can be classified.

This paper presents the problem of unsupervised learning structure and parameter adaptive pattern recognition. Different near optimal solutions that alleviate the infinite or exploding memory requirements of the optimal solution are suggested. The learning capabilities of these methods are compared with that of the optimal supervised scheme by presenting an example.