Multi-branch neural networks with branch control

Takashi Yamashita, Kotaro Hirasawa, Jinglu Hu

研究成果: Conference article


Multi-branch neural networks have been proposed already in order to realize compact networks. It uses some branches between nodes, and this can improve the learning and generalization ability of the networks. In this paper, Branch Control is proposed on the multi-branch neural networks to further enhance the learning and generalization ability of the networks. Branch Control is to adjust the values of the signals on the branches depending on the network inputs using an additional branch control network. It has been clarified from simulation results of a function approximation problem that multi-branch neural networks with Branch Control could be improved more than that without Branch Control.

ジャーナルProceedings of the IEEE International Conference on Systems, Man and Cybernetics
出版物ステータスPublished - 2003 11 24
イベントSystem Security and Assurance - Washington, DC, United States
継続期間: 2003 10 52003 10 8


ASJC Scopus subject areas

  • Control and Systems Engineering
  • Hardware and Architecture