APPLYING SELF-CONFIDENCE IN MULTI-LABEL CLASSIFICATION TO MODEL TRAINING
A computer model is trained to classify regions of a space (e.g., a pixel of an image or a voxel of a point cloud) according to a multi-label classification. To improve the model's accuracy, the model's self-confidence is determined with respect to its own predictions of regions in a train...
Saved in:
Main Authors | , , , |
---|---|
Format | Patent |
Language | English |
Published |
17.03.2022
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Abstract | A computer model is trained to classify regions of a space (e.g., a pixel of an image or a voxel of a point cloud) according to a multi-label classification. To improve the model's accuracy, the model's self-confidence is determined with respect to its own predictions of regions in a training space. The self-confidence is determined based on the class predictions, such as a difference between the highest-predicted class and a second-highest-predicted class. When these are similar, it may reflect areas for potential improvement by focusing training on these low-confidence areas. Additional training may be performed by including modified training data in subsequent training iterations that focuses on low-confidence areas. As another example, additional training may be performed using the self-confidence to modify a classification loss used to refine parameters of the model. |
---|---|
AbstractList | A computer model is trained to classify regions of a space (e.g., a pixel of an image or a voxel of a point cloud) according to a multi-label classification. To improve the model's accuracy, the model's self-confidence is determined with respect to its own predictions of regions in a training space. The self-confidence is determined based on the class predictions, such as a difference between the highest-predicted class and a second-highest-predicted class. When these are similar, it may reflect areas for potential improvement by focusing training on these low-confidence areas. Additional training may be performed by including modified training data in subsequent training iterations that focuses on low-confidence areas. As another example, additional training may be performed using the self-confidence to modify a classification loss used to refine parameters of the model. |
Author | Omer, Om Ji Thyagharajan, Anirud Laddha, Prashant Ummenhofer, Benjamin |
Author_xml | – fullname: Thyagharajan, Anirud – fullname: Laddha, Prashant – fullname: Ummenhofer, Benjamin – fullname: Omer, Om Ji |
BookMark | eNrjYmDJy89L5WTwcAwI8In09HNXCHb1cdN19vdz83Rx9XN2VfD0U_AN9Qnx1PVxdHL1UXD2cQwO9nTzdHYM8fT3UwjxV_D1dwGKhwQ5evoB9fMwsKYl5hSn8kJpbgZlN9cQZw_d1IL8-NTigsTk1LzUkvjQYCMDIyMDCxNjQwNHQ2PiVAEA12svYQ |
ContentType | Patent |
DBID | EVB |
DatabaseName | esp@cenet |
DatabaseTitleList | |
Database_xml | – sequence: 1 dbid: EVB name: esp@cenet url: http://worldwide.espacenet.com/singleLineSearch?locale=en_EP sourceTypes: Open Access Repository |
DeliveryMethod | fulltext_linktorsrc |
Discipline | Medicine Chemistry Sciences Physics |
ExternalDocumentID | US2022084310A1 |
GroupedDBID | EVB |
ID | FETCH-epo_espacenet_US2022084310A13 |
IEDL.DBID | EVB |
IngestDate | Fri Jul 19 13:05:17 EDT 2024 |
IsOpenAccess | true |
IsPeerReviewed | false |
IsScholarly | false |
Language | English |
LinkModel | DirectLink |
MergedId | FETCHMERGED-epo_espacenet_US2022084310A13 |
Notes | Application Number: US202117534558 |
OpenAccessLink | https://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20220317&DB=EPODOC&CC=US&NR=2022084310A1 |
ParticipantIDs | epo_espacenet_US2022084310A1 |
PublicationCentury | 2000 |
PublicationDate | 20220317 |
PublicationDateYYYYMMDD | 2022-03-17 |
PublicationDate_xml | – month: 03 year: 2022 text: 20220317 day: 17 |
PublicationDecade | 2020 |
PublicationYear | 2022 |
RelatedCompanies | Intel Corporation |
RelatedCompanies_xml | – name: Intel Corporation |
Score | 3.3824098 |
Snippet | A computer model is trained to classify regions of a space (e.g., a pixel of an image or a voxel of a point cloud) according to a multi-label classification.... |
SourceID | epo |
SourceType | Open Access Repository |
SubjectTerms | CALCULATING COMPUTING COUNTING PHYSICS |
Title | APPLYING SELF-CONFIDENCE IN MULTI-LABEL CLASSIFICATION TO MODEL TRAINING |
URI | https://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20220317&DB=EPODOC&locale=&CC=US&NR=2022084310A1 |
hasFullText | 1 |
inHoldings | 1 |
isFullTextHit | |
isPrint | |
link | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwfV3dS8MwED_G_HzTqfgxpaD0LejabM0ehnRp6yr9YmtlPo21piBIN1zFf99L7XRPe8yFHMmR-0hyvwvAnc4yXQIWiUFFl1BhzEmqpRrpYyDHWNYRb1QChf2gN0ro87Q7bcDHGgtT1Qn9roojokZlqO9lZa-X_5dYVpVbubpP35G0eHTigaXWp2NNwz1qqNZwYEehFXKV80EyUYPxbx9Db_lg4llpRwbSstK-_TKUuJTlplNxjmA3Qn5FeQwNUbTggK__XmvBvl8_ebdgr8rRzFZIrPVwdQIjM4q8Vzd4Uia25xAeBo5rycsixQ0UP_Fil3jm0PYU7ploLp0aLqzEoeKHFtLjsekGOP4Ubh075iOCk5v9yWKWTDZXop9Bs1gU4hwUhoFPznJq9EROM2rMe_Nc6lfaz2k_Y-wC2ts4XW7vvoJD2ZTJVx2jDc3y80tcozcu05tKiD9lB4XZ |
link.rule.ids | 230,309,786,891,25594,76906 |
linkProvider | European Patent Office |
linkToHtml | http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwfV1LT8JAEJ4QfOBNUeMDtYmmt41SFrociCl92Or2EWgNnAitbWJiCpEa_76ztSgnrjPZye5m57U73yzAXYclHQFYJCpNu4Sm6pzESqyQPgZyjCXt9I0KoLDr9eyIPk-6kxp8rLEwZZ_Q77I5ImpUgvpelPZ6-X-JZZS1lav7-B1Ji0crHBhylR0rCp5RVTaGAzPwDV-XdX0QjWVv9Mtj6C0fNMyVdlTRn1cET69DgUtZbjoV6xB2A5SXF0dQS_MmNPT132tN2HerJ-8m7JU1mskKiZUero7B1oKATx3vSRqb3CK671mOIS6LJMeT3IiHDuHa0OSSzjU0l1YFF5ZCX3J9A-nhSHM8HH8Ct5YZ6jbByc3-9mIWjTdX0jmFer7I0zOQGAY-Gcuo2kszmlB13ptnQr_ifkb7CWPn0Nom6WI7-wYadujyGXe8l0s4ECxRiNVWW1AvPr_SK_TMRXxdbugP0GWIxg |
openUrl | ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Apatent&rft.title=APPLYING+SELF-CONFIDENCE+IN+MULTI-LABEL+CLASSIFICATION+TO+MODEL+TRAINING&rft.inventor=Thyagharajan%2C+Anirud&rft.inventor=Laddha%2C+Prashant&rft.inventor=Ummenhofer%2C+Benjamin&rft.inventor=Omer%2C+Om+Ji&rft.date=2022-03-17&rft.externalDBID=A1&rft.externalDocID=US2022084310A1 |