MASKED MODEL TRAINING OF A PREDICTION NETWORK
In some embodiments, a method receives a first sequence of inputs for processing via a sub-model of a plurality of sub-model. The plurality of sub-models are part of a main model. An input in the sequence of inputs is masked with a masked value to generate a second sequence of inputs. The method pro...
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Patent |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | MAO, Xianghui XIE, Xiaohui ZHAO, Pengyu XU, Chunxu |
| description | In some embodiments, a method receives a first sequence of inputs for processing via a sub-model of a plurality of sub-model. The plurality of sub-models are part of a main model. An input in the sequence of inputs is masked with a masked value to generate a second sequence of inputs. The method processes the second sequence of inputs using the sub-model to generate a sequence of features that correspond to the second sequence of inputs and processes the sequence of features to generate a first output. The first output is processed to generate a second output of the main model. The sub-model is trained based on a feature in the sequence of features that corresponds to the masked input and the second output. |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_US2023019564A1</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>US2023019564A1</sourcerecordid><originalsourceid>FETCH-epo_espacenet_US2023019564A13</originalsourceid><addsrcrecordid>eNrjZND1dQz2dnVR8PV3cfVRCAly9PTz9HNX8HdTcFQICHJ18XQO8fT3U_BzDQn3D_LmYWBNS8wpTuWF0twMym6uIc4euqkF-fGpxQWJyal5qSXxocFGBkbGBoaWpmYmjobGxKkCADylJjs</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>MASKED MODEL TRAINING OF A PREDICTION NETWORK</title><source>esp@cenet</source><creator>MAO, Xianghui ; XIE, Xiaohui ; ZHAO, Pengyu ; XU, Chunxu</creator><creatorcontrib>MAO, Xianghui ; XIE, Xiaohui ; ZHAO, Pengyu ; XU, Chunxu</creatorcontrib><description>In some embodiments, a method receives a first sequence of inputs for processing via a sub-model of a plurality of sub-model. The plurality of sub-models are part of a main model. An input in the sequence of inputs is masked with a masked value to generate a second sequence of inputs. The method processes the second sequence of inputs using the sub-model to generate a sequence of features that correspond to the second sequence of inputs and processes the sequence of features to generate a first output. The first output is processed to generate a second output of the main model. The sub-model is trained based on a feature in the sequence of features that corresponds to the masked input and the second output.</description><language>eng</language><subject>CALCULATING ; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS ; COMPUTING ; COUNTING ; ELECTRIC COMMUNICATION TECHNIQUE ; ELECTRICITY ; PHYSICS ; PICTORIAL COMMUNICATION, e.g. TELEVISION</subject><creationdate>2023</creationdate><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20230119&DB=EPODOC&CC=US&NR=2023019564A1$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,780,885,25564,76547</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20230119&DB=EPODOC&CC=US&NR=2023019564A1$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>MAO, Xianghui</creatorcontrib><creatorcontrib>XIE, Xiaohui</creatorcontrib><creatorcontrib>ZHAO, Pengyu</creatorcontrib><creatorcontrib>XU, Chunxu</creatorcontrib><title>MASKED MODEL TRAINING OF A PREDICTION NETWORK</title><description>In some embodiments, a method receives a first sequence of inputs for processing via a sub-model of a plurality of sub-model. The plurality of sub-models are part of a main model. An input in the sequence of inputs is masked with a masked value to generate a second sequence of inputs. The method processes the second sequence of inputs using the sub-model to generate a sequence of features that correspond to the second sequence of inputs and processes the sequence of features to generate a first output. The first output is processed to generate a second output of the main model. The sub-model is trained based on a feature in the sequence of features that corresponds to the masked input and the second output.</description><subject>CALCULATING</subject><subject>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</subject><subject>COMPUTING</subject><subject>COUNTING</subject><subject>ELECTRIC COMMUNICATION TECHNIQUE</subject><subject>ELECTRICITY</subject><subject>PHYSICS</subject><subject>PICTORIAL COMMUNICATION, e.g. TELEVISION</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2023</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNrjZND1dQz2dnVR8PV3cfVRCAly9PTz9HNX8HdTcFQICHJ18XQO8fT3U_BzDQn3D_LmYWBNS8wpTuWF0twMym6uIc4euqkF-fGpxQWJyal5qSXxocFGBkbGBoaWpmYmjobGxKkCADylJjs</recordid><startdate>20230119</startdate><enddate>20230119</enddate><creator>MAO, Xianghui</creator><creator>XIE, Xiaohui</creator><creator>ZHAO, Pengyu</creator><creator>XU, Chunxu</creator><scope>EVB</scope></search><sort><creationdate>20230119</creationdate><title>MASKED MODEL TRAINING OF A PREDICTION NETWORK</title><author>MAO, Xianghui ; XIE, Xiaohui ; ZHAO, Pengyu ; XU, Chunxu</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_US2023019564A13</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>eng</language><creationdate>2023</creationdate><topic>CALCULATING</topic><topic>COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS</topic><topic>COMPUTING</topic><topic>COUNTING</topic><topic>ELECTRIC COMMUNICATION TECHNIQUE</topic><topic>ELECTRICITY</topic><topic>PHYSICS</topic><topic>PICTORIAL COMMUNICATION, e.g. TELEVISION</topic><toplevel>online_resources</toplevel><creatorcontrib>MAO, Xianghui</creatorcontrib><creatorcontrib>XIE, Xiaohui</creatorcontrib><creatorcontrib>ZHAO, Pengyu</creatorcontrib><creatorcontrib>XU, Chunxu</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>MAO, Xianghui</au><au>XIE, Xiaohui</au><au>ZHAO, Pengyu</au><au>XU, Chunxu</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>MASKED MODEL TRAINING OF A PREDICTION NETWORK</title><date>2023-01-19</date><risdate>2023</risdate><abstract>In some embodiments, a method receives a first sequence of inputs for processing via a sub-model of a plurality of sub-model. The plurality of sub-models are part of a main model. An input in the sequence of inputs is masked with a masked value to generate a second sequence of inputs. The method processes the second sequence of inputs using the sub-model to generate a sequence of features that correspond to the second sequence of inputs and processes the sequence of features to generate a first output. The first output is processed to generate a second output of the main model. The sub-model is trained based on a feature in the sequence of features that corresponds to the masked input and the second output.</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_epo_espacenet_US2023019564A1 |
| source | esp@cenet |
| subjects | CALCULATING COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS COMPUTING COUNTING ELECTRIC COMMUNICATION TECHNIQUE ELECTRICITY PHYSICS PICTORIAL COMMUNICATION, e.g. TELEVISION |
| title | MASKED MODEL TRAINING OF A PREDICTION NETWORK |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-30T09%3A07%3A08IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-epo_EVB&rft_val_fmt=info:ofi/fmt:kev:mtx:patent&rft.genre=patent&rft.au=MAO,%20Xianghui&rft.date=2023-01-19&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3EUS2023019564A1%3C/epo_EVB%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |