Using adaptive logic networks for quick recognition of particles

Abstract

This paper presents research on using adaptive logic networks (a type of neural network) to quickly determine particle types based on momentum and Cherenkov radiation pattern. Two configurations of the network are analyzed. This research also presents new ways of using adaptive logic networks. By taking advantage of the monotonicity property of these networks, more consistent output can be produced and proper unary codes can be generated. Preliminary performance results are presented which indicate that adaptive logic networks are a good candidate for doing particle recognition and other pattern classification tasks requiring great speed. >