The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #92156

Title:The Recall Performance of Multiple Associative Memories
Date: 1992
Abstract:The associative net presents a practical and efficient means for the storage and retrieval of data. Although the recall performance of single-layer associative nets has been well researched [Willshaw et al, 1969 Willshaw, 1971, Buckingham, 1992], little investigation of similar nature has been done on multiple associative nets. The advanced distributed associative memory (ADAM) which was devised by Austin [Austin, 1987] is claimed to have several advantages over the associative net. Several interconnection topologies for the implementation of training and recall simulations on a parallel computer is investigated. An analysis and comparison of the recall performance of ADAM and multiple associative nets using other thresholding strategies notably that of Willshaw [Willshaw et al, 1969], is presented. Simulation results indicate that for a given recall performance criterion there is generally no significant difference between the recall performance of the ADAM net and a two layer nets employing Willshaw thresholding at both layers the ADAM net. The influence of the dimension of intermediate layers on the recall performance is also studied. The effectiveness of alternative thresholding strategies is considered and the performance of single and multilayer associative nets are compared.

[Search These Pages] [DAI Home Page] [Comment]