A neural network architecture, PARADISE, has previously been proposed for general purpose visual pattern recognition. The architecture makes a minimum number of assumptions about the structure of the objects that are to be recognised from the images. This enables it to be applied to a wide variety of pattern and object recognition tasks with a minimal amount of user adaption.

The network derives architectural aspects from the mammalian biological visual system, providing a basis for inherent deformation invariant object recognition. It shares some of the aspects of the Neocognitron and the Feature Recognition Network whilst being simpler and more general purpose. The architecture is structured around a hierarchy of three processing layers, each devoted to a specific processing function; feature extraction, pattern detection and pattern integration.

This paper details the results that were obtained using PARADISE to solve the problems of the classification of face images and hand written character recognition. The results compare favourably with current techniques designed specifically for the two tasks.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £75 +P&P)