This Presidential Young Investigator will continue his research into Associative Memory Networks (AMNs). A wide variety of AMN designs have been proposed, based on different fundamental algorithms. This project will begin with variants of Binary AMNs, and investigate their performance in terms of learning rate, flexibility, and storage capacity. Formal mathematical analysis, using probabilistic concepts, will be stressed. Computer simulations will be used extensively to verify the theory and guide the analysis. Selected AMN designs will then be tested on applications in communications and computer vision, by simulating the designs in computer software. If successful, the designs will then be implemented in special-purpose integrated circuits.
