Risk-Based Semi-Supervised Discriminative Language Modeling for Broadcast Transcription
This paper describes a new method for semi-supervised discriminative language modeling, which is designed to improve the robustness of a discriminative language model (LM) obtained from manually transcribed (labeled) data. The discriminative LM is implemented as a log-linear model, which employs a s...
Gespeichert in:
| Veröffentlicht in: | IEICE Transactions on Information and Systems 2012/11/01, Vol.E95.D(11), pp.2674-2681 |
|---|---|
| Hauptverfasser: | , , , |
| Format: | Artikel |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | 2681 |
|---|---|
| container_issue | 11 |
| container_start_page | 2674 |
| container_title | IEICE Transactions on Information and Systems |
| container_volume | E95.D |
| creator | KOBAYASHI, Akio OKU, Takahiro IMAI, Toru NAKAGAWA, Seiichi |
| description | This paper describes a new method for semi-supervised discriminative language modeling, which is designed to improve the robustness of a discriminative language model (LM) obtained from manually transcribed (labeled) data. The discriminative LM is implemented as a log-linear model, which employs a set of linguistic features derived from word or phoneme sequences. The proposed semi-supervised discriminative modeling is formulated as a multi-objective optimization programming problem (MOP), which consists of two objective functions defined on both labeled lattices and automatic speech recognition (ASR) lattices as unlabeled data. The objectives are coherently designed based on the expected risks that reflect information about word errors for the training data. The model is trained in a discriminative manner and acquired as a solution to the MOP problem. In transcribing Japanese broadcast programs, the proposed method reduced relatively a word error rate by 6.3% compared with that achieved by a conventional trigram LM. |
| doi_str_mv | 10.1587/transinf.E95.D.2674 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_cross</sourceid><recordid>TN_cdi_proquest_miscellaneous_1315698083</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>1315698083</sourcerecordid><originalsourceid>FETCH-LOGICAL-c541t-f4274dcb07ce4234b2b79057c9122beca31fa04c83b1957f87a88840939117883</originalsourceid><addsrcrecordid>eNpdkElP8zAQhi0EEmX5BVxyQfouKR4vsX0EyqoiJBZxtCauUwxp0s9Okfj3JCpUiNNopOed5SHkCOgYpFYnXcQmhaYaXxg5noxZocQWGYESMgdewDYZUQNFriVnu2QvpTdKQTOQI_LyENJ7fobJz7JHvwj542rp40cY-klILoZFaLALHz6bYjNf4dxnd-3M16GZZ1Ubs7PY4sxh6rKn4Yg-sOxC2xyQnQrr5A-_6z55vrx4Or_Op_dXN-en09xJAV1eCabEzJVUOS8YFyUrlaFSOQOMld4hhwqpcJqXYKSqtEKttaCGGwClNd8n_9Zzl7H9v_Kps4v-al_X2Ph2lSxwkIXRVPMe5WvUxTal6Cu77L_D-GmB2kGj_dFoe412YgeNfer4ewEmh3XVIy6kTZQV0gjOB-52zb2lrpe0ATB2wdX-72yAX0s2kHvFaH3DvwDdZ5CE</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>1315698083</pqid></control><display><type>article</type><title>Risk-Based Semi-Supervised Discriminative Language Modeling for Broadcast Transcription</title><source>J-STAGE Free</source><source>EZB-FREE-00999 freely available EZB journals</source><creator>KOBAYASHI, Akio ; OKU, Takahiro ; IMAI, Toru ; NAKAGAWA, Seiichi</creator><creatorcontrib>KOBAYASHI, Akio ; OKU, Takahiro ; IMAI, Toru ; NAKAGAWA, Seiichi</creatorcontrib><description>This paper describes a new method for semi-supervised discriminative language modeling, which is designed to improve the robustness of a discriminative language model (LM) obtained from manually transcribed (labeled) data. The discriminative LM is implemented as a log-linear model, which employs a set of linguistic features derived from word or phoneme sequences. The proposed semi-supervised discriminative modeling is formulated as a multi-objective optimization programming problem (MOP), which consists of two objective functions defined on both labeled lattices and automatic speech recognition (ASR) lattices as unlabeled data. The objectives are coherently designed based on the expected risks that reflect information about word errors for the training data. The model is trained in a discriminative manner and acquired as a solution to the MOP problem. In transcribing Japanese broadcast programs, the proposed method reduced relatively a word error rate by 6.3% compared with that achieved by a conventional trigram LM.</description><identifier>ISSN: 0916-8532</identifier><identifier>EISSN: 1745-1361</identifier><identifier>DOI: 10.1587/transinf.E95.D.2674</identifier><language>eng</language><publisher>Oxford: The Institute of Electronics, Information and Communication Engineers</publisher><subject>Applied sciences ; Artificial intelligence ; Bayes risk minimization ; Broadcasting ; Broadcasting. Videocommunications. Audiovisual ; Computer science; control theory; systems ; discriminative training ; Errors ; Exact sciences and technology ; Information, signal and communications theory ; language modeling ; Lattices ; Linguistics ; Mathematical models ; Miscellaneous ; Optimization ; Programming ; Robustness ; semi-supervised training ; Signal processing ; Speech and sound recognition and synthesis. Linguistics ; Speech processing ; Telecommunications ; Telecommunications and information theory</subject><ispartof>IEICE Transactions on Information and Systems, 2012/11/01, Vol.E95.D(11), pp.2674-2681</ispartof><rights>2012 The Institute of Electronics, Information and Communication Engineers</rights><rights>2015 INIST-CNRS</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><cites>FETCH-LOGICAL-c541t-f4274dcb07ce4234b2b79057c9122beca31fa04c83b1957f87a88840939117883</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>315,781,785,1884,4025,27928,27929,27930</link.rule.ids><backlink>$$Uhttp://pascal-francis.inist.fr/vibad/index.php?action=getRecordDetail&idt=26594334$$DView record in Pascal Francis$$Hfree_for_read</backlink></links><search><creatorcontrib>KOBAYASHI, Akio</creatorcontrib><creatorcontrib>OKU, Takahiro</creatorcontrib><creatorcontrib>IMAI, Toru</creatorcontrib><creatorcontrib>NAKAGAWA, Seiichi</creatorcontrib><title>Risk-Based Semi-Supervised Discriminative Language Modeling for Broadcast Transcription</title><title>IEICE Transactions on Information and Systems</title><addtitle>IEICE Trans. Inf. & Syst.</addtitle><description>This paper describes a new method for semi-supervised discriminative language modeling, which is designed to improve the robustness of a discriminative language model (LM) obtained from manually transcribed (labeled) data. The discriminative LM is implemented as a log-linear model, which employs a set of linguistic features derived from word or phoneme sequences. The proposed semi-supervised discriminative modeling is formulated as a multi-objective optimization programming problem (MOP), which consists of two objective functions defined on both labeled lattices and automatic speech recognition (ASR) lattices as unlabeled data. The objectives are coherently designed based on the expected risks that reflect information about word errors for the training data. The model is trained in a discriminative manner and acquired as a solution to the MOP problem. In transcribing Japanese broadcast programs, the proposed method reduced relatively a word error rate by 6.3% compared with that achieved by a conventional trigram LM.</description><subject>Applied sciences</subject><subject>Artificial intelligence</subject><subject>Bayes risk minimization</subject><subject>Broadcasting</subject><subject>Broadcasting. Videocommunications. Audiovisual</subject><subject>Computer science; control theory; systems</subject><subject>discriminative training</subject><subject>Errors</subject><subject>Exact sciences and technology</subject><subject>Information, signal and communications theory</subject><subject>language modeling</subject><subject>Lattices</subject><subject>Linguistics</subject><subject>Mathematical models</subject><subject>Miscellaneous</subject><subject>Optimization</subject><subject>Programming</subject><subject>Robustness</subject><subject>semi-supervised training</subject><subject>Signal processing</subject><subject>Speech and sound recognition and synthesis. Linguistics</subject><subject>Speech processing</subject><subject>Telecommunications</subject><subject>Telecommunications and information theory</subject><issn>0916-8532</issn><issn>1745-1361</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2012</creationdate><recordtype>article</recordtype><recordid>eNpdkElP8zAQhi0EEmX5BVxyQfouKR4vsX0EyqoiJBZxtCauUwxp0s9Okfj3JCpUiNNopOed5SHkCOgYpFYnXcQmhaYaXxg5noxZocQWGYESMgdewDYZUQNFriVnu2QvpTdKQTOQI_LyENJ7fobJz7JHvwj542rp40cY-klILoZFaLALHz6bYjNf4dxnd-3M16GZZ1Ubs7PY4sxh6rKn4Yg-sOxC2xyQnQrr5A-_6z55vrx4Or_Op_dXN-en09xJAV1eCabEzJVUOS8YFyUrlaFSOQOMld4hhwqpcJqXYKSqtEKttaCGGwClNd8n_9Zzl7H9v_Kps4v-al_X2Ph2lSxwkIXRVPMe5WvUxTal6Cu77L_D-GmB2kGj_dFoe412YgeNfer4ewEmh3XVIy6kTZQV0gjOB-52zb2lrpe0ATB2wdX-72yAX0s2kHvFaH3DvwDdZ5CE</recordid><startdate>2012</startdate><enddate>2012</enddate><creator>KOBAYASHI, Akio</creator><creator>OKU, Takahiro</creator><creator>IMAI, Toru</creator><creator>NAKAGAWA, Seiichi</creator><general>The Institute of Electronics, Information and Communication Engineers</general><general>Oxford University Press</general><scope>IQODW</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>2012</creationdate><title>Risk-Based Semi-Supervised Discriminative Language Modeling for Broadcast Transcription</title><author>KOBAYASHI, Akio ; OKU, Takahiro ; IMAI, Toru ; NAKAGAWA, Seiichi</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c541t-f4274dcb07ce4234b2b79057c9122beca31fa04c83b1957f87a88840939117883</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2012</creationdate><topic>Applied sciences</topic><topic>Artificial intelligence</topic><topic>Bayes risk minimization</topic><topic>Broadcasting</topic><topic>Broadcasting. Videocommunications. Audiovisual</topic><topic>Computer science; control theory; systems</topic><topic>discriminative training</topic><topic>Errors</topic><topic>Exact sciences and technology</topic><topic>Information, signal and communications theory</topic><topic>language modeling</topic><topic>Lattices</topic><topic>Linguistics</topic><topic>Mathematical models</topic><topic>Miscellaneous</topic><topic>Optimization</topic><topic>Programming</topic><topic>Robustness</topic><topic>semi-supervised training</topic><topic>Signal processing</topic><topic>Speech and sound recognition and synthesis. Linguistics</topic><topic>Speech processing</topic><topic>Telecommunications</topic><topic>Telecommunications and information theory</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>KOBAYASHI, Akio</creatorcontrib><creatorcontrib>OKU, Takahiro</creatorcontrib><creatorcontrib>IMAI, Toru</creatorcontrib><creatorcontrib>NAKAGAWA, Seiichi</creatorcontrib><collection>Pascal-Francis</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>IEICE Transactions on Information and Systems</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>KOBAYASHI, Akio</au><au>OKU, Takahiro</au><au>IMAI, Toru</au><au>NAKAGAWA, Seiichi</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Risk-Based Semi-Supervised Discriminative Language Modeling for Broadcast Transcription</atitle><jtitle>IEICE Transactions on Information and Systems</jtitle><addtitle>IEICE Trans. Inf. & Syst.</addtitle><date>2012</date><risdate>2012</risdate><volume>E95.D</volume><issue>11</issue><spage>2674</spage><epage>2681</epage><pages>2674-2681</pages><issn>0916-8532</issn><eissn>1745-1361</eissn><abstract>This paper describes a new method for semi-supervised discriminative language modeling, which is designed to improve the robustness of a discriminative language model (LM) obtained from manually transcribed (labeled) data. The discriminative LM is implemented as a log-linear model, which employs a set of linguistic features derived from word or phoneme sequences. The proposed semi-supervised discriminative modeling is formulated as a multi-objective optimization programming problem (MOP), which consists of two objective functions defined on both labeled lattices and automatic speech recognition (ASR) lattices as unlabeled data. The objectives are coherently designed based on the expected risks that reflect information about word errors for the training data. The model is trained in a discriminative manner and acquired as a solution to the MOP problem. In transcribing Japanese broadcast programs, the proposed method reduced relatively a word error rate by 6.3% compared with that achieved by a conventional trigram LM.</abstract><cop>Oxford</cop><pub>The Institute of Electronics, Information and Communication Engineers</pub><doi>10.1587/transinf.E95.D.2674</doi><tpages>8</tpages><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext |
| identifier | ISSN: 0916-8532 |
| ispartof | IEICE Transactions on Information and Systems, 2012/11/01, Vol.E95.D(11), pp.2674-2681 |
| issn | 0916-8532 1745-1361 |
| language | eng |
| recordid | cdi_proquest_miscellaneous_1315698083 |
| source | J-STAGE Free; EZB-FREE-00999 freely available EZB journals |
| subjects | Applied sciences Artificial intelligence Bayes risk minimization Broadcasting Broadcasting. Videocommunications. Audiovisual Computer science control theory systems discriminative training Errors Exact sciences and technology Information, signal and communications theory language modeling Lattices Linguistics Mathematical models Miscellaneous Optimization Programming Robustness semi-supervised training Signal processing Speech and sound recognition and synthesis. Linguistics Speech processing Telecommunications Telecommunications and information theory |
| title | Risk-Based Semi-Supervised Discriminative Language Modeling for Broadcast Transcription |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-12T23%3A25%3A19IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_cross&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Risk-Based%20Semi-Supervised%20Discriminative%20Language%20Modeling%20for%20Broadcast%20Transcription&rft.jtitle=IEICE%20Transactions%20on%20Information%20and%20Systems&rft.au=KOBAYASHI,%20Akio&rft.date=2012&rft.volume=E95.D&rft.issue=11&rft.spage=2674&rft.epage=2681&rft.pages=2674-2681&rft.issn=0916-8532&rft.eissn=1745-1361&rft_id=info:doi/10.1587/transinf.E95.D.2674&rft_dat=%3Cproquest_cross%3E1315698083%3C/proquest_cross%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=1315698083&rft_id=info:pmid/&rfr_iscdi=true |