A sequential dynamic heteroassociative memory for multistep pattern recognition and one-to-many association

• Published in: IEEE transaction on neural networks and learning systems 2006-01, Vol.17 (1), p.59-68
• Main Authors: Chartier, S., Boukadoum, M.
• Format: Article
• Language: English
• Subjects: Algorithms; Applied sciences; Artificial Intelligence; Associative memories; Associative memory; Backpropagation; bidirectional associative memories (BAMs); Computer science; control theory; systems; Computer Systems; Connectionism. Neural networks; Context modeling; Exact sciences and technology; Hebbian theory; Limit-cycles; Magnesium compounds; Models, Neurological; Neural networks; one-to-many association; Pattern recognition; Pattern Recognition, Automated; Psychology; sequence learning; Supervised learning
• ISSN: 1045-9227; 2162-237X; 1941-0093; 2162-2388
• DOI: 10.1109/TNN.2005.860855

Bidirectional associative memories (BAMs) have been widely used for auto and heteroassociative learning. However, few research efforts have addressed the issue of multistep vector pattern recognition. We propose a model that can perform multi step pattern recognition without the need for a special learning algorithm, and with the capacity to learn more than two pattern series in the training set. The model can also learn pattern series of different lengths and, contrarily to previous models, the stimuli can be composed of gray-level images. The paper also shows that by adding an extra autoassociative layer, the model can accomplish one-to-many association, a task that was exclusive to feedforward networks with context units and error backpropagation learning.