Reformulating Distributed Associative Memories for Image Classification

Fabio Ancona, Stefano Rovetta, and Rodolfo Zunino

Dept. of Biophysical and Electronic Engineering (DIBE),
University of Genova
Via all’Opera Pia 11a,16145 Genova, Italy

ABSTRACT

The theoretical model of Distributed Associative Memories
(DAMs) is reformulated by simple algebraic derivations
that make the memory device practically applicable to
high-dimensional, visual data processing. In particular,
the analysis shows that the weight of both the computational
cost for retrieval and the physical memory occupation can be
reduced from N4 to N2, where N is the data dimensionality.
The presented research extends previous results on the model
implementation on parallel machinery, whose ultimate
efficiency is now characterized from a theoretical point
of view, too. 

----
WCNN96, Proceedings of the 1996 World Congress on
Neural Networks, September 1996.