A Hierarchical Neural Network Model for Category Detection
H. Sakaguchi · Progress of Theoretical Physics · 1989
The present paper proposes a neural network model which has an ability of hierarchical categorization. The network is self-organized by a Hebbian learning rule and each cell in the network is differentiated to detect respectively a set of input signals which corresponds to a category. If mutual connections among the cells are suitable in strength and input signals have a hierarchical structure, a hierarchical relation emerges among the detector cells.