DeepLocalize: Fault Localization for Deep Neural Networks

Deep Neural Networks (DNNs) are becoming an integral part of most software systems. Previous work has shown that DNNs have bugs. Unfortunately, existing debugging techniques don't support localizing DNN bugs because of the lack of understanding of model behaviors. The entire DNN model appears a...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Proceedings / International Conference on Software Engineering s. 251 - 262
Hlavní autori: Wardat, Mohammad, Le, Wei, Rajan, Hridesh
Médium: Konferenčný príspevok..
Jazyk:English
Vydavateľské údaje: IEEE 01.05.2021
Predmet:
Benchmark testing
Computer bugs
Debugging
Deep learning bugs
Deep Neural Networks
Fault diagnosis
Fault Location
Neural networks
Numerical models
Program Analysis
ISBN:1665402962, 9781665402965
ISSN:1558-1225
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Abstract Deep Neural Networks (DNNs) are becoming an integral part of most software systems. Previous work has shown that DNNs have bugs. Unfortunately, existing debugging techniques don't support localizing DNN bugs because of the lack of understanding of model behaviors. The entire DNN model appears as a black box. To address these problems, we propose an approach and a tool that automatically determines whether the model is buggy or not, and identifies the root causes for DNN errors. Our key insight is that historic trends in values propagated between layers can be analyzed to identify faults, and also localize faults. To that end, we first enable dynamic analysis of deep learning applications: by converting it into an imperative representation and alternatively using a callback mechanism. Both mechanisms allows us to insert probes that enable dynamic analysis over the traces produced by the DNN while it is being trained on the training data. We then conduct dynamic analysis over the traces to identify the faulty layer or hyperparameter that causes the error. We propose an algorithm for identifying root causes by capturing any numerical error and monitoring the model during training and finding the relevance of every layer/parameter on the DNN outcome. We have collected a benchmark containing 40 buggy models and patches that contain real errors in deep learning applications from Stack Overflow and GitHub. Our benchmark can be used to evaluate automated debugging tools and repair techniques. We have evaluated our approach using this DNN bug-and-patch benchmark, and the results showed that our approach is much more effective than the existing debugging approach used in the state-of-the-practice Keras library. For 34/40 cases, our approach was able to detect faults whereas the best debugging approach provided by Keras detected 32/40 faults. Our approach was able to localize 21/40 bugs whereas Keras did not localize any faults.
AbstractList Deep Neural Networks (DNNs) are becoming an integral part of most software systems. Previous work has shown that DNNs have bugs. Unfortunately, existing debugging techniques don't support localizing DNN bugs because of the lack of understanding of model behaviors. The entire DNN model appears as a black box. To address these problems, we propose an approach and a tool that automatically determines whether the model is buggy or not, and identifies the root causes for DNN errors. Our key insight is that historic trends in values propagated between layers can be analyzed to identify faults, and also localize faults. To that end, we first enable dynamic analysis of deep learning applications: by converting it into an imperative representation and alternatively using a callback mechanism. Both mechanisms allows us to insert probes that enable dynamic analysis over the traces produced by the DNN while it is being trained on the training data. We then conduct dynamic analysis over the traces to identify the faulty layer or hyperparameter that causes the error. We propose an algorithm for identifying root causes by capturing any numerical error and monitoring the model during training and finding the relevance of every layer/parameter on the DNN outcome. We have collected a benchmark containing 40 buggy models and patches that contain real errors in deep learning applications from Stack Overflow and GitHub. Our benchmark can be used to evaluate automated debugging tools and repair techniques. We have evaluated our approach using this DNN bug-and-patch benchmark, and the results showed that our approach is much more effective than the existing debugging approach used in the state-of-the-practice Keras library. For 34/40 cases, our approach was able to detect faults whereas the best debugging approach provided by Keras detected 32/40 faults. Our approach was able to localize 21/40 bugs whereas Keras did not localize any faults.
Author Le, Wei
Rajan, Hridesh
Wardat, Mohammad
Author_xml – sequence: 1
  givenname: Mohammad
  surname: Wardat
  fullname: Wardat, Mohammad
  email: wardat@iastate.edu
  organization: Iowa State University, USA
– sequence: 2
  givenname: Wei
  surname: Le
  fullname: Le, Wei
  email: weile@iastate.edu
  organization: Iowa State University, USA
– sequence: 3
  givenname: Hridesh
  surname: Rajan
  fullname: Rajan, Hridesh
  email: hridesh@iastate.edu
  organization: Iowa State University, USA
BookMark eNotj81Kw1AUhC9Ywbb2CXSRF0g85_4l153EVgvBLqrrcm5yLgRjUpIU0ac3xW7mm4FhYBZi1nYtC3GPkCCCe9jm-7VWDmQiQWICAEpfiQVaazRIZ-VMzNGYLEYpzY1YDUPtQevUIVg9F-6Z-Vh0JTX1Lz9GGzo1Y3TJNNZdG4Wuj86l6I1PPTUTxu-u_xxuxXWgZuDVhUvxsVm_569xsXvZ5k9FTDIzY2w5CwSeAhqvtVelxRCQPTuuUouqqrQO4NVkUjhL6pm1DGVGKMlbtRR3_7s1Mx-Off1F_c_BTefAGvUHIx1KVw
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IH
CBEJK
RIE
RIO
DOI 10.1109/ICSE43902.2021.00034
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan (POP) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE/IET Electronic Library (IEL) (UW System Shared)
IEEE Proceedings Order Plans (POP) 1998-present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Xplore
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Computer Science
EndPage 262
ExternalDocumentID 9402065
Genre orig-research
GrantInformation_xml – fundername: National Science Foundation
  funderid: 10.13039/100000001
GroupedDBID -~X
.4S
.DC
123
23M
29O
5VS
6IE
6IF
6IH
6IK
6IL
6IM
6IN
8US
AAJGR
AAWTH
ABLEC
ADZIZ
AFFNX
ALMA_UNASSIGNED_HOLDINGS
APO
ARCSS
AVWKF
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
EDO
FEDTE
I-F
I07
IEGSK
IJVOP
IPLJI
M43
OCL
RIE
RIL
RIO
RNS
XOL
ID FETCH-LOGICAL-a285t-6e8fa0baf15b44b3c61ff1ebe9ed7613dd44f0b33dd703dd77bee42fc8a12ab63
IEDL.DBID RIE
ISBN 1665402962
9781665402965
ISICitedReferencesCount 79
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000684601800021&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1558-1225
IngestDate Wed Aug 27 02:28:03 EDT 2025
IsPeerReviewed false
IsScholarly true
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-a285t-6e8fa0baf15b44b3c61ff1ebe9ed7613dd44f0b33dd703dd77bee42fc8a12ab63
PageCount 12
ParticipantIDs ieee_primary_9402065
PublicationCentury 2000
PublicationDate 2021-May
PublicationDateYYYYMMDD 2021-05-01
PublicationDate_xml – month: 05
  year: 2021
  text: 2021-May
PublicationDecade 2020
PublicationTitle Proceedings / International Conference on Software Engineering
PublicationTitleAbbrev ICSE
PublicationYear 2021
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssib044791064
ssj0006499
Score 2.530428
Snippet Deep Neural Networks (DNNs) are becoming an integral part of most software systems. Previous work has shown that DNNs have bugs. Unfortunately, existing...
SourceID ieee
SourceType Publisher
StartPage 251
SubjectTerms Benchmark testing
Computer bugs
Debugging
Deep learning bugs
Deep Neural Networks
Fault diagnosis
Fault Location
Neural networks
Numerical models
Program Analysis
Title DeepLocalize: Fault Localization for Deep Neural Networks
URI https://ieeexplore.ieee.org/document/9402065
WOSCitedRecordID wos000684601800021&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1La8JAEF5UeujJtlr6Zg89NrrZ7CPp1SotFBHagjfZxywIoqJJD_313V2jpdBLL0k25BBmdvPNbOabD6F7DdRaSWxiuKIJYzwsKZEmVsucaO204S6KTcjxOJ9Oi0kDPRy4MAAQi8-gFy7jv3y7MlXYKusXIdkRvImaUoodV2s_dxiTHvhC6F9_hQWL2pEeLn2W5CdtIHUFpV1CC0HrXk_7Ma85dSkp-i-Dt6EH6UjSomkvNnD5pbwSgWfU_t8rn6DuD4MPTw7YdIoasDxD7b2EA65XdAcVTwDr14Bn8y94xCNVLUpcj6PPsA9qcXgIhzYeauFPsW5820Ufo-H74Dmp1RQSRXNeJgJyp4hWLuWaMZ0ZkTqXeh8WYKUHdWsZc0RnWfBdOEgNwKgzuUqp0iI7R63lagkXCBeZJRzA-WTHMaKNtyUF7wLw4Y0CQy9RJ1hitt41zJjVRrj6-_Y1Og6m3lUR3qBWuangFh2Zz3K-3dxFL38D8eKhcw
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1LTwIxEG4QTfSECsa3PXh0odtt9-EVIRCRkIgJN9LHNCEhQHh48NfbloIx8eJld7vZw2am3W-mO998CD1KoFpnREeKCxoxxt2SSuNIyywnUhqpuPFiE1m_n49GxaCEnvZcGADwxWdQd5f-X76eq43bKmsULtlJ-QE6dMpZga21mz2MZRb6XPAfvsMp8-qRFjBtnmSnraN1Oa1dQouUhm5PuzEPrLqYFI1u871lYdrTtGhc9y1cfmmveOhpV_730qeo9sPhw4M9Op2hEszOUWUn4oDDmq6i4gVg0XOINvmCZ9wWm-kah7H3GrZhLXYPYdfIQ0ztyVeOr2roo90aNjtR0FOIBM35OkohN4JIYWIuGZOJSmNjYuvFAnRmYV1rxgyRSeK85w6ZBGDUqFzEVMg0uUDl2XwGlwgXiSYcwNh0xzAilbUlBesCsAGOAEWvUNVZYrzYtswYByNc_337AR13hm-9ca_bf71BJ87s25rCW1ReLzdwh47U53qyWt57j38Dk66kvA
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%3Abook&rft.genre=proceeding&rft.title=Proceedings+%2F+International+Conference+on+Software+Engineering&rft.atitle=DeepLocalize%3A+Fault+Localization+for+Deep+Neural+Networks&rft.au=Wardat%2C+Mohammad&rft.au=Le%2C+Wei&rft.au=Rajan%2C+Hridesh&rft.date=2021-05-01&rft.pub=IEEE&rft.isbn=9781665402965&rft.issn=1558-1225&rft.spage=251&rft.epage=262&rft_id=info:doi/10.1109%2FICSE43902.2021.00034&rft.externalDocID=9402065
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1558-1225&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1558-1225&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1558-1225&client=summon