Most algorithms for solving AI problems are computationally very demanding. The current trend in computer architectures toward parallel machines offers the chance to exploit much more computational power than ever before and to design algorithms which would not have been feasible on sequential machines.
The intent of this course is to bring together ideas from computer architects, AI researchers and application engineers to study the problem of parallel processing, distributed processing, and multi-threading for artificial intelligence. The course will provide a review of common parallel architectures and programming techniques, and will focus on methods of parallelizing fundamental areas of AI.
Hands-on experience parallelizing AI algorithms will be given by designing, testing and refining parallel AI algorithms using the 128-processor nCUBE 2, the 8-processor Sequent, a SIMD simulator, a threads package, and a distributed network of PCs.