Title:
学習装置、パラメータの作成方法、ニューラルネットワーク及びこれを用いた情報処理装置
Document Type and Number:
Japanese Patent JP7316771
Kind Code:
B2
Abstract:
To more effectively perform learning of a neural network.SOLUTION: A learning device performs learning of a neural network. The learning device sets teacher data for each of two or more different outputs from the neural network, corresponding to a single input data for learning. The learning device performs the learning of the neural network on the basis of an error between each of the two or more different outputs and the teacher data corresponding to the outputs, obtained by inputting the input data for learning in the neural network. The neural network after learning provides the two or more different outputs corresponding to the input data, and an integration result of the two or more different outputs indicates a result of recognition processing for the input data.SELECTED DRAWING: Figure 1
Inventors:
Koichi Tanji
Atsushi Nogami
Yusuke Mitarai
Atsushi Nogami
Yusuke Mitarai
Application Number:
JP2018170893A
Publication Date:
July 28, 2023
Filing Date:
September 12, 2018
Export Citation:
Assignee:
Canon Inc
International Classes:
G06N3/08; G06T7/00
Domestic Patent References:
| JP2016031746A | ||||
| JP2012022535A |
Other References:
LIU Yu et al.,Learning Relaxed Deep Supervision for Better Edge Detection,2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) [online],2016年,pp. 231-240,[2022年8月23日検索], インターネット
久保田 涼介 ほか ,全層畳み込みニューラルネットワークを用いた透明物体の輪郭抽出,電子情報通信学会技術研究報告 PRMU2018-10,vol.118, No.35,2018年05月10日,pp.41-46
久保田 涼介 ほか ,全層畳み込みニューラルネットワークを用いた透明物体の輪郭抽出,電子情報通信学会技術研究報告 PRMU2018-10,vol.118, No.35,2018年05月10日,pp.41-46
Attorney, Agent or Firm:
Patent Attorney Corporation Otsuka International Patent Office