WEKO3
アイテム
{"_buckets": {"deposit": "086eba84-3e1a-462e-88fb-e7e89c2cd90d"}, "_deposit": {"created_by": 28, "id": "946", "owners": [28], "pid": {"revision_id": 0, "type": "depid", "value": "946"}, "status": "published"}, "_oai": {"id": "oai:oist.repo.nii.ac.jp:00000946", "sets": ["78"]}, "author_link": ["5194", "5192", "5191", "5193"], "item_10001_biblio_info_7": {"attribute_name": "Bibliographic Information", "attribute_value_mlt": [{"bibliographicIssueDates": {"bibliographicIssueDate": "2018-11-28", "bibliographicIssueDateType": "Issued"}, "bibliographicPageStart": "78", "bibliographicVolumeNumber": "12", "bibliographic_titles": [{}, {"bibliographic_title": "Frontiers in Neurorobotics", "bibliographic_titleLang": "en"}]}]}, "item_10001_creator_3": {"attribute_name": "Author", "attribute_type": "creator", "attribute_value_mlt": [{"creatorNames": [{"creatorName": "Parisi, German I."}], "nameIdentifiers": [{"nameIdentifier": "5191", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Tani, Jun"}], "nameIdentifiers": [{"nameIdentifier": "5192", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Weber, Cornelius"}], "nameIdentifiers": [{"nameIdentifier": "5193", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Wermter, Stefan"}], "nameIdentifiers": [{"nameIdentifier": "5194", "nameIdentifierScheme": "WEKO"}]}]}, "item_10001_description_5": {"attribute_name": "Abstract", "attribute_value_mlt": [{"subitem_description": "Artificial autonomous agents and robots interacting in complex environments are required to continually acquire and fine-tune knowledge over sustained periods of time. The ability to learn from continuous streams of information is referred to as lifelong learning and represents a long-standing challenge for neural network models due to catastrophic forgetting in which novel sensory experience interferes with existing representations and leads to abrupt decreases in the performance on previously acquired knowledge. Computational models of lifelong learning typically alleviate catastrophic forgetting in experimental scenarios with given datasets of static images and limited complexity, thereby differing significantly from the conditions artificial agents are exposed to. In more natural settings, sequential information may become progressively available over time and access to previous experience may be restricted. Therefore, specialized neural network mechanisms are required that adapt to novel sequential experience while preventing disruptive interference with existing representations. In this paper, we propose a dual-memory self-organizing architecture for lifelong learning scenarios. The architecture comprises two growing recurrent networks with the complementary tasks of learning object instances (episodic memory) and categories (semantic memory). Both growing networks can expand in response to novel sensory experience: the episodic memory learns fine-grained spatiotemporal representations of object instances in an unsupervised fashion while the semantic memory uses task-relevant signals to regulate structural plasticity levels and develop more compact representations from episodic experience. For the consolidation of knowledge in the absence of external sensory input, the episodic memory periodically replays trajectories of neural reactivations. We evaluate the proposed model on the CORe50 benchmark dataset for continuous object recognition, showing that we significantly outperform current methods of lifelong learning in three different incremental learning scenarios.", "subitem_description_type": "Other"}]}, "item_10001_publisher_8": {"attribute_name": "Publisher", "attribute_value_mlt": [{"subitem_publisher": "Frontiers Media"}]}, "item_10001_relation_13": {"attribute_name": "PubMedNo.", "attribute_value_mlt": [{"subitem_relation_type": "isIdenticalTo", "subitem_relation_type_id": {"subitem_relation_type_id_text": "info:pmid/30546302", "subitem_relation_type_select": "PMID"}}]}, "item_10001_relation_14": {"attribute_name": "DOI", "attribute_value_mlt": [{"subitem_relation_type": "isIdenticalTo", "subitem_relation_type_id": {"subitem_relation_type_id_text": "info:doi/10.3389/fnbot.2018.00078", "subitem_relation_type_select": "DOI"}}]}, "item_10001_relation_16": {"attribute_name": "情報源", "attribute_value_mlt": [{"subitem_relation_name": [{"subitem_relation_name_text": "https://creativecommons.org/licenses/by/4.0/"}]}]}, "item_10001_relation_17": {"attribute_name": "Related site", "attribute_value_mlt": [{"subitem_relation_type_id": {"subitem_relation_type_id_text": "https://www.frontiersin.org/articles/10.3389/fnbot.2018.00078/full", "subitem_relation_type_select": "URI"}}]}, "item_10001_rights_15": {"attribute_name": "Rights", "attribute_value_mlt": [{"subitem_rights": "© 2018 Parisi, Tani, Weber and Wermter. "}]}, "item_10001_source_id_9": {"attribute_name": "ISSN", "attribute_value_mlt": [{"subitem_source_identifier": "1662-5218", "subitem_source_identifier_type": "ISSN"}]}, "item_10001_version_type_20": {"attribute_name": "Author\u0027s flag", "attribute_value_mlt": [{"subitem_version_resource": "http://purl.org/coar/version/c_970fb48d4fbd8a85", "subitem_version_type": "VoR"}]}, "item_files": {"attribute_name": "ファイル情報", "attribute_type": "file", "attribute_value_mlt": [{"accessrole": "open_date", "date": [{"dateType": "Available", "dateValue": "2019-06-18"}], "displaytype": "detail", "download_preview_message": "", "file_order": 0, "filename": "Parisi-2018-Lifelong Learning of Spatiotempora.pdf", "filesize": [{"value": "1.8 MB"}], "format": "application/pdf", "future_date_message": "", "is_thumbnail": false, "licensefree": "Creative Commons\nAttribution 4.0 International\n(https://creativecommons.org/licenses/by/4.0/)", "licensetype": "license_free", "mimetype": "application/pdf", "size": 1800000.0, "url": {"label": "Parisi-2018-Lifelong Learning of Spatiotempora", "url": "https://oist.repo.nii.ac.jp/record/946/files/Parisi-2018-Lifelong Learning of Spatiotempora.pdf"}, "version_id": "d75183c0-e8c7-4886-85e1-ee5d1bdb69c0"}]}, "item_language": {"attribute_name": "言語", "attribute_value_mlt": [{"subitem_language": "eng"}]}, "item_resource_type": {"attribute_name": "資源タイプ", "attribute_value_mlt": [{"resourcetype": "journal article", "resourceuri": "http://purl.org/coar/resource_type/c_6501"}]}, "item_title": "Lifelong Learning of Spatiotemporal Representations With Dual-Memory Recurrent Self-Organization", "item_titles": {"attribute_name": "タイトル", "attribute_value_mlt": [{"subitem_title": "Lifelong Learning of Spatiotemporal Representations With Dual-Memory Recurrent Self-Organization", "subitem_title_language": "en"}]}, "item_type_id": "10001", "owner": "28", "path": ["78"], "permalink_uri": "https://oist.repo.nii.ac.jp/records/946", "pubdate": {"attribute_name": "公開日", "attribute_value": "2019-06-18"}, "publish_date": "2019-06-18", "publish_status": "0", "recid": "946", "relation": {}, "relation_version_is_last": true, "title": ["Lifelong Learning of Spatiotemporal Representations With Dual-Memory Recurrent Self-Organization"], "weko_shared_id": 28}
Lifelong Learning of Spatiotemporal Representations With Dual-Memory Recurrent Self-Organization
https://oist.repo.nii.ac.jp/records/946
https://oist.repo.nii.ac.jp/records/946d4e568cb-d2c5-4351-bcaa-4c6c5d39f488
名前 / ファイル | ライセンス | アクション |
---|---|---|
Parisi-2018-Lifelong Learning of Spatiotempora (1.8 MB)
|
Creative Commons
Attribution 4.0 International (https://creativecommons.org/licenses/by/4.0/) |
Item type | 学術雑誌論文 / Journal Article(1) | |||||
---|---|---|---|---|---|---|
公開日 | 2019-06-18 | |||||
タイトル | ||||||
言語 | en | |||||
タイトル | Lifelong Learning of Spatiotemporal Representations With Dual-Memory Recurrent Self-Organization | |||||
言語 | ||||||
言語 | eng | |||||
資源タイプ | ||||||
資源タイプ識別子 | http://purl.org/coar/resource_type/c_6501 | |||||
資源タイプ | journal article | |||||
著者(英) |
Parisi, German I.
× Parisi, German I.× Tani, Jun× Weber, Cornelius× Wermter, Stefan |
|||||
書誌情報 |
en : Frontiers in Neurorobotics 巻 12, p. 78, 発行日 2018-11-28 |
|||||
抄録 | ||||||
内容記述タイプ | Other | |||||
内容記述 | Artificial autonomous agents and robots interacting in complex environments are required to continually acquire and fine-tune knowledge over sustained periods of time. The ability to learn from continuous streams of information is referred to as lifelong learning and represents a long-standing challenge for neural network models due to catastrophic forgetting in which novel sensory experience interferes with existing representations and leads to abrupt decreases in the performance on previously acquired knowledge. Computational models of lifelong learning typically alleviate catastrophic forgetting in experimental scenarios with given datasets of static images and limited complexity, thereby differing significantly from the conditions artificial agents are exposed to. In more natural settings, sequential information may become progressively available over time and access to previous experience may be restricted. Therefore, specialized neural network mechanisms are required that adapt to novel sequential experience while preventing disruptive interference with existing representations. In this paper, we propose a dual-memory self-organizing architecture for lifelong learning scenarios. The architecture comprises two growing recurrent networks with the complementary tasks of learning object instances (episodic memory) and categories (semantic memory). Both growing networks can expand in response to novel sensory experience: the episodic memory learns fine-grained spatiotemporal representations of object instances in an unsupervised fashion while the semantic memory uses task-relevant signals to regulate structural plasticity levels and develop more compact representations from episodic experience. For the consolidation of knowledge in the absence of external sensory input, the episodic memory periodically replays trajectories of neural reactivations. We evaluate the proposed model on the CORe50 benchmark dataset for continuous object recognition, showing that we significantly outperform current methods of lifelong learning in three different incremental learning scenarios. | |||||
出版者 | ||||||
出版者 | Frontiers Media | |||||
ISSN | ||||||
収録物識別子タイプ | ISSN | |||||
収録物識別子 | 1662-5218 | |||||
PubMed番号 | ||||||
関連タイプ | isIdenticalTo | |||||
識別子タイプ | PMID | |||||
関連識別子 | info:pmid/30546302 | |||||
DOI | ||||||
関連タイプ | isIdenticalTo | |||||
識別子タイプ | DOI | |||||
関連識別子 | info:doi/10.3389/fnbot.2018.00078 | |||||
権利 | ||||||
権利情報 | © 2018 Parisi, Tani, Weber and Wermter. | |||||
情報源 | ||||||
関連名称 | https://creativecommons.org/licenses/by/4.0/ | |||||
関連サイト | ||||||
識別子タイプ | URI | |||||
関連識別子 | https://www.frontiersin.org/articles/10.3389/fnbot.2018.00078/full | |||||
著者版フラグ | ||||||
出版タイプ | VoR | |||||
出版タイプResource | http://purl.org/coar/version/c_970fb48d4fbd8a85 |