The correlational associative memory realizes Hebbian learning

✍ Scribed by N.B. Karayiannis; A.N. Venetsanopoulos

• Publisher: Elsevier Science
• Year: 1988
• Tongue: English
• Weight: 64 KB
• Volume: 1
• Category: Article
• ISSN: 0893-6080
• DOI: 10.1016/0893-6080(88)90224-9

✦ Synopsis

CAM) was originally proposed as an ad hoc scheme in the early seventies. In the early eighties, the validity of this type of associative memory was justified on the basis of rather intuitive arguments and using asynchronous network dynamics. Since then, a series of interesting new results on the subject have been published. Most of the papers, study the performance of the CAM, with particular emphasis on the crucial problem of the capacity of the memory. This paper presents a derivation of the CAM based upon a certain type of learning, called Hebbian learning. Such a derivation provides a better understanding of this type of learning and also underlines the limitations of this type of associative memory. On the other hand, this analysis may be a motivation for the development of even more sophisticated and efficient schemes of associative memory.