Enriching Graph Information for Pedestrian Behavior Learning

• Item type: SIG Technical Reports(1)
• 公開日: 2019-03-02
• タイトル: Enriching Graph Information for Pedestrian Behavior Learning
• タイトル / 言語: en
• タイトル: Enriching Graph Information for Pedestrian Behavior Learning
• 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: National Institute of Advanced Industrial Science and Technology
• 著者所属: National Institute of Advanced Industrial Science and Technology
• 著者所属: 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
• 著者所属(英): en
• 著者所属(英): 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
• 著者名: Nahum, Alvarez; Chenyi, Zhuang; Itsuki, Noda
• 著者名(英): Nahum, Alvarez; Chenyi, Zhuang; Itsuki, Noda
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: Many aspects in planning and prediction in real environments could benefit from behavior learning techniques, with pedestrian simulation for city and business planning being one of them. We developed a method for behavior learning called Contextual Action Multiple Policy Inverse Reinforcement Learning (CAMP-IRL) that allows effective pedestrian behavior simulation. However we noticed some limitations when identifying important relationships between locations in the city map. In this paper we present a graph enrichment technique devised to solve this issue and similar ones. The technique works by applying a number of transformations to the map features in order to permeate with feature information the nodes that are indirectly related to those features. Our tests showed promising results, improving the performance of the original CAMP-IRL method and opening new paths to explore.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: Many aspects in planning and prediction in real environments could benefit from behavior learning techniques, with pedestrian simulation for city and business planning being one of them. We developed a method for behavior learning called Contextual Action Multiple Policy Inverse Reinforcement Learning (CAMP-IRL) that allows effective pedestrian behavior simulation. However we noticed some limitations when identifying important relationships between locations in the city map. In this paper we present a graph enrichment technique devised to solve this issue and similar ones. The technique works by applying a number of transformations to the map features in order to permeate with feature information the nodes that are indirectly related to those features. Our tests showed promising results, improving the performance of the original CAMP-IRL method and opening new paths to explore.
• 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: AA11135936
• 書誌情報: 研究報告知能システム（ICS）; 巻 2019-ICS-194, 号 7, p. 1-6, 発行日 2019-03-02
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-885X
• 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