Trends in knowledge base processing using optical techniques

A review is presented of recent research efforts, along with a discussion of the relative merits and limitations of using optics as an implementation technology. To a large extent, past efforts have focused on representing knowledge using matrix-like formalisms and designing system architectures based on optical inner-product processors (such as matrix-vector multipliers) and optical correlators. Actual implementations are impeded primarily by the limitations of current spatial light modulators. New directions include the use of symbolic substitution and neural network ideas