Dalarna University's logo and link to the university's website

du.sePublikasjoner
Endre søk
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • chicago-author-date
  • chicago-note-bibliography
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf
Interpreting convolutional neural network by joint evaluation of multiple feature maps and an improved NSGA-II algorithm
Högskolan Dalarna, Institutionen för information och teknik, Mikrodataanalys.ORCID-id: 0000-0003-4212-8582
2024 (engelsk)Inngår i: Expert systems with applications, ISSN 0957-4174, E-ISSN 1873-6793, Vol. 255, artikkel-id 124489Artikkel i tidsskrift (Fagfellevurdert) Published
Abstract [en]

The ’black box’ characteristics of Convolutional Neural Networks (CNNs) present significant risks to their application scenarios, such as reliability, security, and division of responsibilities. Addressing the interpretability of CNN emerges as an urgent and critical issue in the field of machine learning. Recent research on CNN interpretability has either yielded unstable or inconsistent interpretations, or produced coarse-scale interpretable heatmaps, limiting their applicability in various scenarios. In this work, we propose a novel method of CNNs interpretation by incorporating a joint evaluation of multiple feature maps and employing multi-objective optimization (JE&MOO-CAM). Firstly, a method of joint evaluation for all feature maps is proposed to preserve the complete object instances and improve the overall activation values. Secondly, an interpretation method of CNNs under the MOO framework is proposed to avoid the instability and inconsistency of interpretation. Finally, the operators of selection, crossover, and mutation, along with the method of population initialization in NSGA-II, are redesigned to properly express the characteristics of CNNs. The experimental results, including both qualitative and quantitative assessments along with a sanity check conducted on three classic CNN models—VGG16, AlexNet, and ResNet50—demonstrate the superior performance of the proposed JE&MOO-CAM model. This model not only accurately pinpoints the instances within the image requiring explanation but also preserves the integrity of these instances to the greatest extent possible. These capabilities signify that JE&MOO-CAM surpasses six other leading state-of-the-art methods across four established evaluation criteria.

sted, utgiver, år, opplag, sider
2024. Vol. 255, artikkel-id 124489
Emneord [en]
Black box, Convolutional neural network, Interpretability, Feature map, Multi-objective optimization
HSV kategori
Identifikatorer
URN: urn:nbn:se:du-48949DOI: 10.1016/j.eswa.2024.124489ISI: 001262094900001Scopus ID: 2-s2.0-85196977450OAI: oai:DiVA.org:du-48949DiVA, id: diva2:1881795
Tilgjengelig fra: 2024-07-03 Laget: 2024-07-03 Sist oppdatert: 2024-09-18bibliografisk kontrollert

Open Access i DiVA

fulltext(43185 kB)41 nedlastinger
Filinformasjon
Fil FULLTEXT01.pdfFilstørrelse 43185 kBChecksum SHA-512
62d337f6aacc050650fe1fe3a84f3198d05331aa4c76a80bd84305f7b9bf3d3644c01a6e55dd7edf002d9b9234f1edb38e2bdaaa5d1bf1c0139844aa517661c4
Type fulltextMimetype application/pdf

Andre lenker

Forlagets fulltekstScopus

Person

Han, Mengjie

Søk i DiVA

Av forfatter/redaktør
Wang, ZhenwuHan, Mengjie
Av organisasjonen
I samme tidsskrift
Expert systems with applications

Søk utenfor DiVA

GoogleGoogle Scholar
Totalt: 41 nedlastinger
Antall nedlastinger er summen av alle nedlastinger av alle fulltekster. Det kan for eksempel være tidligere versjoner som er ikke lenger tilgjengelige

doi
urn-nbn

Altmetric

doi
urn-nbn
Totalt: 323 treff
RefereraExporteraLink to record
Permanent link

Direct link
Referera
Referensformat
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • chicago-author-date
  • chicago-note-bibliography
  • Annet format
Fler format
Språk
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Annet språk
Fler språk
Utmatningsformat
  • html
  • text
  • asciidoc
  • rtf