Node-perturbation Learning applied for Soft-committee machine

• Item type: SIG Technical Reports(1)
• 公開日: 2019-12-04
• タイトル: Node-perturbation Learning applied for Soft-committee machine
• タイトル / 言語: en
• タイトル: Node-perturbation Learning applied for Soft-committee machine
• 言語: eng
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: College of Industrial Technology, Nihon University
• 著者所属: Graduate School of Informatics, Nagoya University
• 著者所属: Graduate School of Sciences, The University of Tokyo／Graduate School of Frontier Sciences, The University of Tokyo
• 著者所属(英): en
• 著者所属(英): College of Industrial Technology, Nihon University
• 著者所属(英): en
• 著者所属(英): Graduate School of Informatics, Nagoya University
• 著者所属(英): en
• 著者所属(英): Graduate School of Sciences, The University of Tokyo / Graduate School of Frontier Sciences, The University of Tokyo
• 著者名: Kazuyuki, Hara; Kentaro, Katahira; Masato, Okada
• 著者名(英): Kazuyuki, Hara; Kentaro, Katahira; Masato, Okada
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Node-perturbation learning is a stochastic gradient descent method for neural networks. It estimates the gradient of the error surface by calculating the change in error between the perturbed output and the non-perturbed output. Node-perturbation can be applied to problems where the objective function is not defined. We explore the application of node perturbation learning to a multilayer neural network called a soft committee machine and analyze the dynamic properties of the learning process. We conduct computer simulations to show the validity of the proposed method.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Node-perturbation learning is a stochastic gradient descent method for neural networks. It estimates the gradient of the error surface by calculating the change in error between the perturbed output and the non-perturbed output. Node-perturbation can be applied to problems where the objective function is not defined. We explore the application of node perturbation learning to a multilayer neural network called a soft committee machine and analyze the dynamic properties of the learning process. We conduct computer simulations to show the validity of the proposed method.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AN10505667
• 書誌情報: 研究報告数理モデル化と問題解決（MPS）; 巻 2019-MPS-126, 号 2, p. 1-6, 発行日 2019-12-04
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8833
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会