Pattern Matching Co-Processor for Advanced Reasoning Systems

Real-time systems are being called upon to do more complex and inter-related tasks than ever before. Expert systems and other rule-based system are a standard AI technique used to codify human expertise and reasoning capability n a form that a computer can use. However, large systems are typically too slow for many realtime applications. Reticular Systems proposes to solve this problem by developing a VLSI co-processor that will speed up such expert system processing by a factor of 10-1000 times. This increase in speed will open up many new real-time applications that require the intelligence of expert systems. Moreover, the similarity between the low level pattern matching of expert systems and the join process in large relational database searches indicate that this co-processor could be utilized in many database applications where getting the query results in as short a time as possible is critical. Reticular Systems experience in developing intelligent systems (both software and hardware) gives us the unique opportunity to develop and exploit this technology in both commercial and military markets.