Development of pattern recognition systems usually begins with selection of interesting features for resolving the problem. Later appearance of new interesting features must be taken into account in order to be used by the same system without developing the system again completely. The network configuration called Follower-Associative Configuration is analyzed in classification task. This network with multidendritic connectivity presents representations generated in multiple dendritic fascicles. The associated dendritic representations of the same layer can be selected as diverse processing pathways by means of simple modulation mechanism.
展开▼
