Applications of Distributed Mining Techniques For Knowledge Discovery in Dispersed Sensory Data

This paper describes an InferAgent approach, its implementation, and its commercial deployment. The deployed application generates inductive models from dispersed sensory data via application of a distributed data mining algorithm. It facilitates information fusion, composite discrimination, and optimal sensor querying. In this approach, global inductive models of region-of-interests in images are learned as classification rules via tree induction from distributed data. Agents, in a collaborative fashion, generate partial trees and communicate the temporary results among them in the form of indices to the data points. The process is terminated when a final tree is induced. This communication mechanism does not involve any data transfers, and in addition, a compression approach is used to reduce the communication bandwidth of data index transfers. The Combat Vision Laboratory of Lockheed Martin Missiles and Fire Control (LMMFC) is the first deployment site for the InferAgent system. The user payoff has been identified in the data ownership preserving distributed mining mechanism and the capability to discover querying patterns from dispersed sensory data.