DOMINO: Domain-aware Loss for Deep Learning Calibration

Deep learning has achieved the state-of-the-art performance across medical imaging tasks; however, model calibration is often not considered. Uncalibrated models are potentially dangerous in high-risk applications since the user does not know when they will fail. Therefore, this paper proposes a nov...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Stolte, Skylar E, Volle, Kyle, Indahlastari, Aprinda, Albizu, Alejandro, Woods, Adam J, Brink, Kevin, Hale, Matthew, Fang, Ruogu
Format:	Artikel
Sprache:	eng
Schlagworte:	Computer Science - Artificial Intelligence Computer Science - Computer Vision and Pattern Recognition Computer Science - Learning
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	Stolte, Skylar E Volle, Kyle Indahlastari, Aprinda Albizu, Alejandro Woods, Adam J Brink, Kevin Hale, Matthew Fang, Ruogu
description	Deep learning has achieved the state-of-the-art performance across medical imaging tasks; however, model calibration is often not considered. Uncalibrated models are potentially dangerous in high-risk applications since the user does not know when they will fail. Therefore, this paper proposes a novel domain-aware loss function to calibrate deep learning models. The proposed loss function applies a class-wise penalty based on the similarity between classes within a given target domain. Thus, the approach improves the calibration while also ensuring that the model makes less risky errors even when incorrect. The code for this software is available at https://github.com/lab-smile/DOMINO.
doi_str_mv	10.48550/arxiv.2302.05142
format	Article
fullrecord	<record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2302_05142</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2302_05142</sourcerecordid><originalsourceid>FETCH-LOGICAL-a672-9c66a51f560671871e75bc7ae22d19d79338ddbed80ba298e68a5e3bc4bf49c93</originalsourceid><addsrcrecordid>eNotj7tuwjAUQL0wVMAHdMI_kOBH_GJDSWmRAlnYo-v4BlmCBDlVC39flXY629E5hLxylhdWKbaGdI9fuZBM5EzxQrwQUzWH_bHZ0Gq8Qhwy-IaEtB6nifZjohXijdYIaYjDmZZwiT7BZxyHBZn1cJlw-c85Oe3eTuVHVjfv-3JbZ6CNyFynNSjeK8204dZwNMp3BlCIwF0wTkobgsdgmQfhLGoLCqXvCt8XrnNyTlZ_2md5e0vxCunR_h60zwP5A0d7P0Q</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>DOMINO: Domain-aware Loss for Deep Learning Calibration</title><source>arXiv.org</source><creator>Stolte, Skylar E ; Volle, Kyle ; Indahlastari, Aprinda ; Albizu, Alejandro ; Woods, Adam J ; Brink, Kevin ; Hale, Matthew ; Fang, Ruogu</creator><creatorcontrib>Stolte, Skylar E ; Volle, Kyle ; Indahlastari, Aprinda ; Albizu, Alejandro ; Woods, Adam J ; Brink, Kevin ; Hale, Matthew ; Fang, Ruogu</creatorcontrib><description>Deep learning has achieved the state-of-the-art performance across medical imaging tasks; however, model calibration is often not considered. Uncalibrated models are potentially dangerous in high-risk applications since the user does not know when they will fail. Therefore, this paper proposes a novel domain-aware loss function to calibrate deep learning models. The proposed loss function applies a class-wise penalty based on the similarity between classes within a given target domain. Thus, the approach improves the calibration while also ensuring that the model makes less risky errors even when incorrect. The code for this software is available at https://github.com/lab-smile/DOMINO.</description><identifier>DOI: 10.48550/arxiv.2302.05142</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Computer Vision and Pattern Recognition ; Computer Science - Learning</subject><creationdate>2023-02</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2302.05142$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2302.05142$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Stolte, Skylar E</creatorcontrib><creatorcontrib>Volle, Kyle</creatorcontrib><creatorcontrib>Indahlastari, Aprinda</creatorcontrib><creatorcontrib>Albizu, Alejandro</creatorcontrib><creatorcontrib>Woods, Adam J</creatorcontrib><creatorcontrib>Brink, Kevin</creatorcontrib><creatorcontrib>Hale, Matthew</creatorcontrib><creatorcontrib>Fang, Ruogu</creatorcontrib><title>DOMINO: Domain-aware Loss for Deep Learning Calibration</title><description>Deep learning has achieved the state-of-the-art performance across medical imaging tasks; however, model calibration is often not considered. Uncalibrated models are potentially dangerous in high-risk applications since the user does not know when they will fail. Therefore, this paper proposes a novel domain-aware loss function to calibrate deep learning models. The proposed loss function applies a class-wise penalty based on the similarity between classes within a given target domain. Thus, the approach improves the calibration while also ensuring that the model makes less risky errors even when incorrect. The code for this software is available at https://github.com/lab-smile/DOMINO.</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Computer Vision and Pattern Recognition</subject><subject>Computer Science - Learning</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj7tuwjAUQL0wVMAHdMI_kOBH_GJDSWmRAlnYo-v4BlmCBDlVC39flXY629E5hLxylhdWKbaGdI9fuZBM5EzxQrwQUzWH_bHZ0Gq8Qhwy-IaEtB6nifZjohXijdYIaYjDmZZwiT7BZxyHBZn1cJlw-c85Oe3eTuVHVjfv-3JbZ6CNyFynNSjeK8204dZwNMp3BlCIwF0wTkobgsdgmQfhLGoLCqXvCt8XrnNyTlZ_2md5e0vxCunR_h60zwP5A0d7P0Q</recordid><startdate>20230210</startdate><enddate>20230210</enddate><creator>Stolte, Skylar E</creator><creator>Volle, Kyle</creator><creator>Indahlastari, Aprinda</creator><creator>Albizu, Alejandro</creator><creator>Woods, Adam J</creator><creator>Brink, Kevin</creator><creator>Hale, Matthew</creator><creator>Fang, Ruogu</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20230210</creationdate><title>DOMINO: Domain-aware Loss for Deep Learning Calibration</title><author>Stolte, Skylar E ; Volle, Kyle ; Indahlastari, Aprinda ; Albizu, Alejandro ; Woods, Adam J ; Brink, Kevin ; Hale, Matthew ; Fang, Ruogu</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a672-9c66a51f560671871e75bc7ae22d19d79338ddbed80ba298e68a5e3bc4bf49c93</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Computer Vision and Pattern Recognition</topic><topic>Computer Science - Learning</topic><toplevel>online_resources</toplevel><creatorcontrib>Stolte, Skylar E</creatorcontrib><creatorcontrib>Volle, Kyle</creatorcontrib><creatorcontrib>Indahlastari, Aprinda</creatorcontrib><creatorcontrib>Albizu, Alejandro</creatorcontrib><creatorcontrib>Woods, Adam J</creatorcontrib><creatorcontrib>Brink, Kevin</creatorcontrib><creatorcontrib>Hale, Matthew</creatorcontrib><creatorcontrib>Fang, Ruogu</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Stolte, Skylar E</au><au>Volle, Kyle</au><au>Indahlastari, Aprinda</au><au>Albizu, Alejandro</au><au>Woods, Adam J</au><au>Brink, Kevin</au><au>Hale, Matthew</au><au>Fang, Ruogu</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>DOMINO: Domain-aware Loss for Deep Learning Calibration</atitle><date>2023-02-10</date><risdate>2023</risdate><abstract>Deep learning has achieved the state-of-the-art performance across medical imaging tasks; however, model calibration is often not considered. Uncalibrated models are potentially dangerous in high-risk applications since the user does not know when they will fail. Therefore, this paper proposes a novel domain-aware loss function to calibrate deep learning models. The proposed loss function applies a class-wise penalty based on the similarity between classes within a given target domain. Thus, the approach improves the calibration while also ensuring that the model makes less risky errors even when incorrect. The code for this software is available at https://github.com/lab-smile/DOMINO.</abstract><doi>10.48550/arxiv.2302.05142</doi><oa>free_for_read</oa></addata></record>
fulltext	fulltext_linktorsrc
identifier	DOI: 10.48550/arxiv.2302.05142
ispartof
issn
language	eng
recordid	cdi_arxiv_primary_2302_05142
source	arXiv.org
subjects	Computer Science - Artificial Intelligence Computer Science - Computer Vision and Pattern Recognition Computer Science - Learning
title	DOMINO: Domain-aware Loss for Deep Learning Calibration
url	https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-02T00%3A33%3A52IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=DOMINO:%20Domain-aware%20Loss%20for%20Deep%20Learning%20Calibration&rft.au=Stolte,%20Skylar%20E&rft.date=2023-02-10&rft_id=info:doi/10.48550/arxiv.2302.05142&rft_dat=%3Carxiv_GOX%3E2302_05142%3C/arxiv_GOX%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