紀要論文

Study on Chunking Mechanisms for a 3-layered Associative Memory and Recall Model

eng

24321192

https://doi.org/10.15002/00021927

法政大学大学院情報科学研究科

法政大学大学院紀要. 情報科学研究科編

14

1

6

2019-03-31

Version of Record

Memory models are important components that support AI computational models to learn and remember things. Usually, computers operate using an address-based memory for storage where each memory item resides in a unique location. This study proposes a 3-layered associative memory model based on chunking mechanisms of the brain to store knowledge in form of association between entities. The human brain is a cognitive model that derives information from sensory data like vision, auditory, and touch, associates different patterns to create knowledge and uses chunking mechanisms to package the acquired knowledge into manageable entities. Through Chunking, each item held in the STM is a singular entity (chunk) containing more associations (knowledge) in it. Using chunking mechanisms of the brain, AMR model can store knowledge in manner that enables faster response to stimuli. To represent knowledge and semantic relations effectively, the hyper structure (concept) defined in denotational mathematics is employed. A network of concepts is maintained in a cognitive knowledge base that continually evolves as knowledge accumulates. The chunking mechanisms used in this study are goal-oriented chunking and automatic chunking.

Article

AA12746425