Abstract

Selecting a closest or "distinguished" vector from a set of vectors is a fundamental operation in many applications, including self-organizing neural networks. The closest-vector-selection (CVS) operation determines the exemplar vector (drawn from a set of vectors) that is closest to an arbitrary input vector presented to the system. CVS systems are useful in a number of applications, such as self-organizing neural networks for vector quantization. Applications include communication and radar target identification. We present the design of an optical system that performs the CVS operation. The design uses a gray wedge and a Hughes liquid-crystal light valve for an optical matrix-vector operation and in various logic operations.

© 1990 Optical Society of America