37 - Hybrid Expert Systems: An Approach to Combining Neural Computation and Rule-Based Reasoning

Abstract

This chapter examines an approach that integrates neural computation and rule-based reasoning, or the hybrid systems. This integration is actively applied in artificial intelligence and cognitive sciences, such as linguistic theory, natural language processing, and expert systems. The opportunity of employing neural techniques in expert systems is often suggested on the ground that the learning, generalization, fault, and noise tolerance capacities of neural networks can alleviate well-known shortcomings of symbolic problem solvers, such as brittleness in front of incomplete or noisy data, no increase in performance with experience, and time-consuming knowledge acquisition. This chapter explores neurosymbolic integration for rule-based expert systems in connection with automatic data acquisition, rule processing, and explanation. At the periphery of expert systems, sensory processing by neural nets is coupled to rule-based reasoning in order to perform a data acquisition task involving the deployment of expert knowledge and heuristic problem solving. The reaction times of rule-based systems are dramatically reduced by the use of a neurally inspired, parallel inference engine. Informative user interactions with expert systems are achieved by coupling symbolic and neurally supported, pictorial explanation. The relative significance of these aspects of neurosymbolic integration is enhanced by pointing to limitations of neural techniques for automatic knowledge acquisition and robust problem solving in expert systems. These uses of neural nets may often jeopardize an expert system's reliability and reduce its transparency to the user.