Auto-identification of a drug seller utilizing a specialized supervised neural network

Abstract

This chapter addresses the problem of pattern recognition in drug enforcement: how an ensemble of artificial adaptive systems is able to distinguish between different ranks of drug dealers using only the limited features available from felons at the moment of arrest. A subset of the most promising features are selected from the set of all possible features utilizing the TWIST system; a collection of classical back-propagation artificial networks are used for pattern recognition tasks, and a new metaclassifier algorithm is shown to optimize the final intelligent classification.