Lightwave Associative Memory That Memorizes and Recalls Information Depending on Optical-Carrier Frequency

Abstract

This chapter presents a lightwave neural network that learns behavior depending on optical frequency. In general, a neural network learns or self-organizes adaptively to environment. Because of this adaptability, we sometimes have difficulty in controlling the neural network at will. The lightwave network to be presented here learns a certain processing at a certain optical-carrier frequency, and another processing at another frequency, to possess both the adaptability and the controllability. From a different viewpoint, we can consider this compatibility as a mean of multiplexing of behavior in the optical frequency domain. The frequency-domain multiplexing provides the neural network, which is characterized by distributedness and parallelism, with a new dimension of massive parallelism utilizing the vast optical frequency bandwidth. In this sense, the network presents a novel direction in optical information processing hardware.