{"updated":"2025-01-19T21:29:09.028072+00:00","links":{},"created":"2025-01-19T01:03:42.482203+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00199987","sets":["6164:6165:6210:9955"]},"path":["9955"],"owner":"44499","recid":"199987","title":["An Attempt to Improve Generalization Performance in Reinforcement Learning with Deterministic World Models and WGANs"],"pubdate":{"attribute_name":"公開日","attribute_value":"2019-11-01"},"_buckets":{"deposit":"075fbe5a-fd4b-405b-b6af-32071e139574"},"_deposit":{"id":"199987","pid":{"type":"depid","value":"199987","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"An Attempt to Improve Generalization Performance in Reinforcement Learning with Deterministic World Models and WGANs","author_link":["485558","485560","485559","485557"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"An Attempt to Improve Generalization Performance in Reinforcement Learning with Deterministic World Models and WGANs"},{"subitem_title":"An Attempt to Improve Generalization Performance in Reinforcement Learning with Deterministic World Models and WGANs","subitem_title_language":"en"}]},"item_type_id":"18","publish_date":"2019-11-01","item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"eng"}]},"item_18_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"Department of Information and Communication Engineering, The University of Tokyo"},{"subitem_text_value":"Department of Information and Communication Engineering, The School of Information Science and Technology, The Uni-versity of Tokyo"}]},"item_18_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Department of Information and Communication Engineering, The University of Tokyo","subitem_text_language":"en"},{"subitem_text_value":"Department of Information and Communication Engineering, The School of Information Science and Technology, The Uni-versity of Tokyo","subitem_text_language":"en"}]},"item_publisher":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会","subitem_publisher_language":"ja"}]},"publish_status":"0","weko_shared_id":-1,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/199987/files/IPSJ-GPWS2019024.pdf","label":"IPSJ-GPWS2019024.pdf"},"date":[{"dateType":"Available","dateValue":"2019-11-01"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-GPWS2019024.pdf","filesize":[{"value":"3.0 MB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"5"},{"tax":["include_tax"],"price":"0","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"18"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"24fde8e2-d1bb-457d-84ee-de1393fc45f6","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2019 by the Information Processing Society of Japan"}]},"item_18_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Tianshuai, Yu"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Yoshimasa, Tsuruoka"}],"nameIdentifiers":[{}]}]},"item_18_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Tianshuai, Yu","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Yoshimasa, Tsuruoka","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_5794","resourcetype":"conference paper"}]},"item_18_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"Significant progress has been made in the field of Reinforcement Learning (RL) in recent years. Using artificial neural networks, researchers are able to train agents that can play video games as well as or even better than human experts. However, it is common that the same environments are used in both training phases and testing phases, which results in agents’ failure to generalize to other environments. In this work, we propose a method in which environment models and generative models are used to generate virtual game levels so as to improve the generalization performance of RL agents. We conducted experiments using a fully-observable deterministic discrete maze game in order to test the proposed method. However, the proposed method failed to converge during training because our environmnet model was not able to predict the future of unseen levels accurately.","subitem_description_type":"Other"}]},"item_18_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"Significant progress has been made in the field of Reinforcement Learning (RL) in recent years. Using artificial neural networks, researchers are able to train agents that can play video games as well as or even better than human experts. However, it is common that the same environments are used in both training phases and testing phases, which results in agents’ failure to generalize to other environments. In this work, we propose a method in which environment models and generative models are used to generate virtual game levels so as to improve the generalization performance of RL agents. We conducted experiments using a fully-observable deterministic discrete maze game in order to test the proposed method. However, the proposed method failed to converge during training because our environmnet model was not able to predict the future of unseen levels accurately.","subitem_description_type":"Other"}]},"item_18_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"154","bibliographic_titles":[{"bibliographic_title":"ゲームプログラミングワークショップ2019論文集"}],"bibliographicPageStart":"150","bibliographicIssueDates":{"bibliographicIssueDate":"2019-11-01","bibliographicIssueDateType":"Issued"},"bibliographicVolumeNumber":"2019"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"id":199987}