Conjoint Computing: Integrating Numeric, Symbolic and Neural Computing

This paper describes an ongoing research effort to develop a single, unified computational paradigm for conjoint computing, which integrates concepts from symbolic processing, numeric processing and neural network technologies. The result of this research effort will be a new methodology for synthesizing intelligent systems. Neither the symbolic nor neural computational paradigms is sufficiently powerful to implement truly intelligent systems given the current state of these technologies. However, by combining these technologies, it will be possible to build systems that behave intelligently, i.e., operate in real time, exhibit adaptive, goal-oriented problem-solving skills, tolerate errors, exploit large amounts of knowledge, use symbols and abstractions, and learn from the environment. Combining symbolic and neural network technologies and drawing on insights developed from the study of biological systems results in systems which do not exhibit the "brittleness" of current symbolic processors yet are able to plan, reason, and perform other cognitive processing tasks. We describe a search effort that is developing the conceptual, architectural, hardware and software framework required for conjoint computing. Our preliminary research has resulted in a multi-layered computational model which is described herein.