We propose a new 2-layer self-organizing neural network which can store information distributedly. The proposed network has two phases for the information processing: the storage phase and the recall phase. In the storage phase, the proposed network represents information distributedly by plural excited neurons using the proposed multi-winners competitive dynamics. And then, the weights between two layers are learned using error correction learning. In the recall phase, the trained proposed network can recall the stored pattern which is the closest to a presented pattern. We carried out computer simulations to confirm the validity of the proposed network.
|ジャーナル||Proceedings of the IEEE International Conference on Systems, Man and Cybernetics|
|出版ステータス||Published - 1997 12月 1|
|イベント||Proceedings of the 1997 IEEE International Conference on Systems, Man, and Cybernetics. Part 3 (of 5) - Orlando, FL, USA|
継続期間: 1997 10月 12 → 1997 10月 15
ASJC Scopus subject areas