A data parallel implementation of an intelligent reasoning system
Kevin Livingston, Jennifer Seitzer · 2002
We present an implementation of a data parallel system. A sequential knowledge-based deductive and inductive system, INDED, is transformed into a parallel system. In this parallel system the learning algorithm, the fundamental component of the induction engine, is distributed among many processors. The parallel system is implemented with a master node and several worker nodes. The master node is responsible for coordinating the activity of the worker nodes, and organizing the overall learning process. All the worker nodes share the processing of the basic induction algorithms and report their results to the master node. The goal of the data parallel system is to produce, more efficiently, rules that are equal to or better than those produced by the serial system. In this paper, we present the architecture of the parallel version of INDED, and comparison results involving execution speeds and quality of generated rules of the new parallel system to those of the serial system.