Appoaches to filtering in parallel

Abstract

We have shown in this chapter how filtering can be performed in parallel. We started with a simple scheme for implementing filtering and filtering with tagging on multiprocessors. Then, we introduced how an arbitrary binary constraint satisfaction problem, i.e., a set of 0-ary and binary dynamic constraints, can be transformed into a connectionist network consisting of 0-nodes and 2-nodes. 2-nodes are initialized with either 1 or 0, depending on whether they correspond to an admissible value combination or an inconsistent one; 0-nodes are initialized with a Gödel number, representing the possible paths of solutions.