In another project, we examine basic processes related to automatic programming, learning, and inference theory. Specifically, we are interested in mechanisms that can receive samples of a behavior and then create a general representation of the behavior which accounts for the given samples and all "similar" cases. For example, the system might receive samples of the input-output behavior for a computer program and then generate the program. A number of models have been studied including grammatical inference systems, program generators, signature table systems and others. We emphasize the development of systems that can be shown to learn well-defined classes of behaviors and whose convergence characterizations can be understood. We seek both new algorithms for doing such inference and general theory to explain the phenomena.