TAM (Topographic Attentive Mapping) network is a biologically-motivated neural network with Gabor function type receptive elds. However, the structure of receptive elds is a mono-layer, and there is a lack of performance for rotating images. In this paper, we formulate a new TAM network with multilayer structure of extensive receptive elds. We also show the usefulness of TAM network using some examples of character recognition.
Isao Hayashi, James R. Williamson