スマートフォン音声データを用いた歯磨き行動評価のためのニューラルネットワーク構造の検討

• Item type: SIG Technical Reports(1)
• 公開日: 2016-11-30
• タイトル: スマートフォン音声データを用いた歯磨き行動評価のためのニューラルネットワーク構造の検討
• タイトル / 言語: en
• タイトル: Preliminary Investigation on Using Deep Learning to Evaluate Toothbrushing Performance with Smartphone Audio
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: MBLセッション3
• 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: 大阪大学大学院情報科学研究科
• 著者所属: 大阪大学大学院情報科学研究科
• 著者所属(英): en
• 著者所属(英): Osaka University
• 著者所属(英): en
• 著者所属(英): Osaka University
• 著者名: Joseph, Korpela; 前川, 卓也
• 著者名(英): Joseph, Korpela; Takuya, Maekawa
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Previous methods for skill assessment using ubiquitous computing have relied on a two-tiered approach that consists of an initial activity recognition process followed by a skill assessment process that uses the results of activity recognition as input. The intermediate activity recognition process used in those methods increases the burden placed on researchers when designing and training their skill assessment system. In this paper, we propose a method for skill assessment that removes the need for an intermediate activity recognition process. We exploit the ability of deep neural networks to extract high-level features from input data to allow us to run a skill assessment model that takes raw sensor data as input. We evaluate our method on the task of toothbrushing performance evaluation, and show that deep neural networks have the potential to compete with more traditional skill assessment systems.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Previous methods for skill assessment using ubiquitous computing have relied on a two-tiered approach that consists of an initial activity recognition process followed by a skill assessment process that uses the results of activity recognition as input. The intermediate activity recognition process used in those methods increases the burden placed on researchers when designing and training their skill assessment system. In this paper, we propose a method for skill assessment that removes the need for an intermediate activity recognition process. We exploit the ability of deep neural networks to extract high-level features from input data to allow us to run a skill assessment model that takes raw sensor data as input. We evaluate our method on the task of toothbrushing performance evaluation, and show that deep neural networks have the potential to compete with more traditional skill assessment systems.
• 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: AA11515904
• 書誌情報: 研究報告高度交通システムとスマートコミュニティ（ITS）; 巻 2016-ITS-67, 号 25, p. 1-8, 発行日 2016-11-30
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8965
• 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