Adaptation of Ray, a distributed framework for machine learning, to MPI-based environment

• Item type: SIG Technical Reports(1)
• 公開日: 2018-07-23
• タイトル: Adaptation of Ray, a distributed framework for machine learning, to MPI-based environment
• タイトル / 言語: en
• タイトル: Adaptation of Ray, a distributed framework for machine learning, to MPI-based environment
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 機械学習・ニューラルネットワーク
• 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: University of Tsukuba／National Institute of Advanced Industrial Science and Technology
• 著者所属: National Institute of Advanced Industrial Science and Technology／University of Tsukuba
• 著者所属: National Institute of Advanced Industrial Science and Technology／University of Tsukuba
• 著者所属(英): en
• 著者所属(英): University of Tsukuba / National Institute of Advanced Industrial Science and Technology
• 著者所属(英): en
• 著者所属(英): National Institute of Advanced Industrial Science and Technology / University of Tsukuba
• 著者所属(英): en
• 著者所属(英): National Institute of Advanced Industrial Science and Technology / University of Tsukuba
• 著者名: Tianlun, Wang; Yusuke, Tanimura; Hidemoto, Nakada
• 著者名(英): Tianlun, Wang; Yusuke, Tanimura; Hidemoto, Nakada
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Ray is a distributed framework for machine learning that targets reinforcement learning using multiple nodes. While it works well on loosely coupled nodes, it does not take into account the "high-performance computing environment " based on MPI. We modified Ray so that, 1) it works well with the MPI launch mechanism, and 2) it use MPI communication for large data transfer. We evaluated the modified version of Ray on a cluster and confirmed the preliminary performance.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Ray is a distributed framework for machine learning that targets reinforcement learning using multiple nodes. While it works well on loosely coupled nodes, it does not take into account the "high-performance computing environment " based on MPI. We modified Ray so that, 1) it works well with the MPI launch mechanism, and 2) it use MPI communication for large data transfer. We evaluated the modified version of Ray on a cluster and confirmed the preliminary performance.
• 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: AN10096105
• 書誌情報: 研究報告システム・アーキテクチャ（ARC）; 巻 2018-ARC-232, 号 29, p. 1-6, 発行日 2018-07-23
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8574
• 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