Holographic On-Line Learning Machine for Multicategory Classification

A holographic on-line learning machine that is capable of multicategory classification is described. The system exactly implements the single-layer perceptron algorithm in a fully parallel and analog fashion. The performance of the adaptive network is successfully tested for up to 24 characters with different scale and rotation. Also, a compact and robust version of the holographic learning machine is proposed.