Hierarchical Concept Formation in Associative Memory Composed of Neuro-window Elements

✍ Scribed by Hideka Kakeya; Toshiki Kindo

• Publisher: Elsevier Science
• Year: 1996
• Tongue: English
• Weight: 728 KB
• Volume: 9
• Category: Article
• ISSN: 0893-6080
• DOI: 10.1016/0893-6080(96)00030-5

✦ Synopsis

In the present letter, the authors propose a model of hierarchical concept formation in auto-associative memory composed of neuro-window elements. When hierarchically correlated patterns are memorized, each center of the correlated memory patterns can be defined as a concept pattern which represents those similar patterns. It is shown that the concept patterns are formed under a Hebbian learning rule in the neural network where all the neuro-window elements are just mutually connected, and that the memory patterns and the concept patterns in each level of hierarchy can be retrieved selectively by adjusting the parameters of the neuro-window elements. Copyright 1996 Elsevier Science Ltd