WEKO3
アイテム
{"_buckets": {"deposit": "6edd34be-360b-4bdf-85c0-4a81dff4eceb"}, "_deposit": {"created_by": 3, "id": "3086", "owners": [3], "pid": {"revision_id": 0, "type": "depid", "value": "3086"}, "status": "published"}, "_oai": {"id": "oai:repository.ninjal.ac.jp:00003086", "sets": ["324"]}, "author_link": ["10422", "10423", "10424", "10425", "10426"], "control_number": "3086", "item_10001_biblio_info_7": {"attribute_name": "書誌情報", "attribute_value_mlt": [{"bibliographicIssueDates": {"bibliographicIssueDate": "2020-07-09", "bibliographicIssueDateType": "Issued"}, "bibliographicPageEnd": "60", "bibliographicPageStart": "55", "bibliographic_titles": [{"bibliographic_title": "Proceedings of the 5th Workshop on Representation Learning for NLP (RepL4NLP-2020)", "bibliographic_titleLang": "en"}]}]}, "item_10001_description_19": {"attribute_name": "フォーマット", "attribute_value_mlt": [{"subitem_description": "application/pdf", "subitem_description_type": "Other"}]}, "item_10001_description_25": {"attribute_name": "著者所属(英)", "attribute_value_mlt": [{"subitem_description": "Ochanomizu University", "subitem_description_type": "Other"}, {"subitem_description": "Microsoft Research", "subitem_description_type": "Other"}, {"subitem_description": "Kyoto University", "subitem_description_type": "Other"}, {"subitem_description": "National Institute for Japanese Language and Linguistics", "subitem_description_type": "Other"}, {"subitem_description": "Ochanomizu University", "subitem_description_type": "Other"}]}, "item_10001_description_26": {"attribute_name": "抄録(英)", "attribute_value_mlt": [{"subitem_description": "We propose an AdversariaL training algorithm for commonsense InferenCE (ALICE). We apply small perturbations to word embeddings and minimize the resultant adversarial risk to regularize the model. We exploit a novel combination of two different approaches to estimate these perturbations: 1) using the true label and 2) using the model prediction. Without relying on any human-crafted features, knowledge bases, or additional datasets other than the target datasets, our model boosts the fine-tuning performance of RoBERTa, achieving competitive results on multiple reading comprehension datasets that require commonsense inference.", "subitem_description_type": "Other"}]}, "item_10001_identifier_registration": {"attribute_name": "ID登録", "attribute_value_mlt": [{"subitem_identifier_reg_text": "10.15084/00003070", "subitem_identifier_reg_type": "JaLC"}]}, "item_10001_publisher_8": {"attribute_name": "出版者", "attribute_value_mlt": [{"subitem_publisher": "Association for Computational Linguistics"}]}, "item_10001_relation_10": {"attribute_name": "ISBN", "attribute_value_mlt": [{"subitem_relation_type": "isPartOf", "subitem_relation_type_id": {"subitem_relation_type_id_text": "978-1-952148-15-6", "subitem_relation_type_select": "ISBN"}}]}, "item_10001_version_type_20": {"attribute_name": "著者版フラグ", "attribute_value_mlt": [{"subitem_version_resource": "http://purl.org/coar/version/c_970fb48d4fbd8a85", "subitem_version_type": "VoR"}]}, "item_creator": {"attribute_name": "著者", "attribute_type": "creator", "attribute_value_mlt": [{"creatorNames": [{"creatorName": "Pereira, Lis", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "10422", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Liu, Xiaodong", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "10423", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Cheng, Fei", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "10424", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Asahara, Masayuki", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "10425", "nameIdentifierScheme": "WEKO"}]}, {"creatorNames": [{"creatorName": "Kobayashi, Ichiro", "creatorNameLang": "en"}], "nameIdentifiers": [{"nameIdentifier": "10426", "nameIdentifierScheme": "WEKO"}]}]}, "item_files": {"attribute_name": "ファイル情報", "attribute_type": "file", "attribute_value_mlt": [{"accessrole": "open_date", "date": [{"dateType": "Available", "dateValue": "2020-12-17"}], "displaytype": "detail", "download_preview_message": "", "file_order": 0, "filename": "repl4nlp_2020_55.pdf", "filesize": [{"value": "2.0 MB"}], "format": "application/pdf", "future_date_message": "", "is_thumbnail": false, "licensetype": "license_free", "mimetype": "application/pdf", "size": 2000000.0, "url": {"label": "repl4nlp_2020_55.pdf", "url": "https://repository.ninjal.ac.jp/record/3086/files/repl4nlp_2020_55.pdf"}, "version_id": "59da4fff-4391-4d4f-9dd2-20fba75ad3a1"}]}, "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": "Adversarial Training for Commonsense Inference", "item_titles": {"attribute_name": "タイトル", "attribute_value_mlt": [{"subitem_title": "Adversarial Training for Commonsense Inference", "subitem_title_language": "en"}]}, "item_type_id": "10001", "owner": "3", "path": ["324"], "permalink_uri": "https://doi.org/10.15084/00003070", "pubdate": {"attribute_name": "PubDate", "attribute_value": "2020-12-18"}, "publish_date": "2020-12-18", "publish_status": "0", "recid": "3086", "relation": {}, "relation_version_is_last": true, "title": ["Adversarial Training for Commonsense Inference"], "weko_shared_id": -1}
Adversarial Training for Commonsense Inference
https://doi.org/10.15084/00003070
https://doi.org/10.15084/00003070c84a62ae-676a-4aac-845a-ff5fa9879bda
名前 / ファイル | ライセンス | アクション |
---|---|---|
![]() |
|
Item type | 学術雑誌論文 / Journal Article(1) | |||||
---|---|---|---|---|---|---|
公開日 | 2020-12-18 | |||||
タイトル | ||||||
言語 | en | |||||
タイトル | Adversarial Training for Commonsense Inference | |||||
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 | |||||
ID登録 | ||||||
ID登録 | 10.15084/00003070 | |||||
ID登録タイプ | JaLC | |||||
著者 |
Pereira, Lis
× Pereira, Lis× Liu, Xiaodong× Cheng, Fei× Asahara, Masayuki× Kobayashi, Ichiro |
|||||
著者所属(英) | ||||||
内容記述タイプ | Other | |||||
内容記述 | Ochanomizu University | |||||
著者所属(英) | ||||||
内容記述タイプ | Other | |||||
内容記述 | Microsoft Research | |||||
著者所属(英) | ||||||
内容記述タイプ | Other | |||||
内容記述 | Kyoto University | |||||
著者所属(英) | ||||||
内容記述タイプ | Other | |||||
内容記述 | National Institute for Japanese Language and Linguistics | |||||
著者所属(英) | ||||||
内容記述タイプ | Other | |||||
内容記述 | Ochanomizu University | |||||
抄録(英) | ||||||
内容記述タイプ | Other | |||||
内容記述 | We propose an AdversariaL training algorithm for commonsense InferenCE (ALICE). We apply small perturbations to word embeddings and minimize the resultant adversarial risk to regularize the model. We exploit a novel combination of two different approaches to estimate these perturbations: 1) using the true label and 2) using the model prediction. Without relying on any human-crafted features, knowledge bases, or additional datasets other than the target datasets, our model boosts the fine-tuning performance of RoBERTa, achieving competitive results on multiple reading comprehension datasets that require commonsense inference. | |||||
出版者 | ||||||
出版者 | Association for Computational Linguistics | |||||
書誌情報 |
en : Proceedings of the 5th Workshop on Representation Learning for NLP (RepL4NLP-2020) p. 55-60, 発行日 2020-07-09 |
|||||
ISBN | ||||||
関連タイプ | isPartOf | |||||
識別子タイプ | ISBN | |||||
関連識別子 | 978-1-952148-15-6 | |||||
フォーマット | ||||||
内容記述タイプ | Other | |||||
内容記述 | application/pdf | |||||
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 |