Matches in SemOpenAlex for { <https://semopenalex.org/work/W3037274327> ?p ?o ?g. }
Showing items 1 to 97 of
97
with 100 items per page.
- W3037274327 abstract "Convolutional neural network (CNN) models for computer vision are powerful but lack explainability in their most basic form. This deficiency remains a key challenge when applying CNNs in important domains. Recent work for explanations through feature importance of approximate linear models has moved from input-level features (pixels or segments) to features from mid-layer feature maps in the guise of concept activation vectors (CAVs). CAVs contain concept-level information and could be learnt via Clustering. In this work, we rethink the ACE algorithm of Ghorbani et~al., proposing an alternative concept-based explanation framework. Based on the requirements of fidelity (approximate models) and interpretability (being meaningful to people), we design measurements and evaluate a range of dimensionality reduction methods for alignment with our framework. We find that non-negative concept activation vectors from non-negative matrix factorization provide superior performance in interpretability and fidelity based on computational and human subject experiments. Our framework provides both local and global concept-level explanations for pre-trained CNN models." @default.
- W3037274327 created "2020-07-02" @default.
- W3037274327 creator A5028824146 @default.
- W3037274327 creator A5031938909 @default.
- W3037274327 creator A5053457984 @default.
- W3037274327 creator A5078824132 @default.
- W3037274327 creator A5084653483 @default.
- W3037274327 date "2020-06-27" @default.
- W3037274327 modified "2023-10-18" @default.
- W3037274327 title "Improving Interpretability of CNN Models Using Non-Negative Concept Activation Vectors." @default.
- W3037274327 cites W1797268635 @default.
- W3037274327 cites W1849277567 @default.
- W3037274327 cites W1902027874 @default.
- W3037274327 cites W1923115158 @default.
- W3037274327 cites W1984314602 @default.
- W3037274327 cites W2113882472 @default.
- W3037274327 cites W2117539524 @default.
- W3037274327 cites W2141708418 @default.
- W3037274327 cites W2194775991 @default.
- W3037274327 cites W2250539671 @default.
- W3037274327 cites W2282821441 @default.
- W3037274327 cites W2332488709 @default.
- W3037274327 cites W2551974706 @default.
- W3037274327 cites W2792641098 @default.
- W3037274327 cites W2895739182 @default.
- W3037274327 cites W2945976633 @default.
- W3037274327 cites W2951306478 @default.
- W3037274327 cites W2963483561 @default.
- W3037274327 cites W2963749936 @default.
- W3037274327 cites W2970030610 @default.
- W3037274327 cites W2971048680 @default.
- W3037274327 hasPublicationYear "2020" @default.
- W3037274327 type Work @default.
- W3037274327 sameAs 3037274327 @default.
- W3037274327 citedByCount "1" @default.
- W3037274327 countsByYear W30372743272021 @default.
- W3037274327 crossrefType "posted-content" @default.
- W3037274327 hasAuthorship W3037274327A5028824146 @default.
- W3037274327 hasAuthorship W3037274327A5031938909 @default.
- W3037274327 hasAuthorship W3037274327A5053457984 @default.
- W3037274327 hasAuthorship W3037274327A5078824132 @default.
- W3037274327 hasAuthorship W3037274327A5084653483 @default.
- W3037274327 hasConcept C111030470 @default.
- W3037274327 hasConcept C119857082 @default.
- W3037274327 hasConcept C138885662 @default.
- W3037274327 hasConcept C153180895 @default.
- W3037274327 hasConcept C154945302 @default.
- W3037274327 hasConcept C2776401178 @default.
- W3037274327 hasConcept C2776459999 @default.
- W3037274327 hasConcept C2781067378 @default.
- W3037274327 hasConcept C41008148 @default.
- W3037274327 hasConcept C41895202 @default.
- W3037274327 hasConcept C70518039 @default.
- W3037274327 hasConcept C73555534 @default.
- W3037274327 hasConcept C76155785 @default.
- W3037274327 hasConcept C81363708 @default.
- W3037274327 hasConceptScore W3037274327C111030470 @default.
- W3037274327 hasConceptScore W3037274327C119857082 @default.
- W3037274327 hasConceptScore W3037274327C138885662 @default.
- W3037274327 hasConceptScore W3037274327C153180895 @default.
- W3037274327 hasConceptScore W3037274327C154945302 @default.
- W3037274327 hasConceptScore W3037274327C2776401178 @default.
- W3037274327 hasConceptScore W3037274327C2776459999 @default.
- W3037274327 hasConceptScore W3037274327C2781067378 @default.
- W3037274327 hasConceptScore W3037274327C41008148 @default.
- W3037274327 hasConceptScore W3037274327C41895202 @default.
- W3037274327 hasConceptScore W3037274327C70518039 @default.
- W3037274327 hasConceptScore W3037274327C73555534 @default.
- W3037274327 hasConceptScore W3037274327C76155785 @default.
- W3037274327 hasConceptScore W3037274327C81363708 @default.
- W3037274327 hasLocation W30372743271 @default.
- W3037274327 hasOpenAccess W3037274327 @default.
- W3037274327 hasPrimaryLocation W30372743271 @default.
- W3037274327 hasRelatedWork W2290300449 @default.
- W3037274327 hasRelatedWork W270432131 @default.
- W3037274327 hasRelatedWork W2796096336 @default.
- W3037274327 hasRelatedWork W2899291531 @default.
- W3037274327 hasRelatedWork W2902376486 @default.
- W3037274327 hasRelatedWork W2947128531 @default.
- W3037274327 hasRelatedWork W3003212594 @default.
- W3037274327 hasRelatedWork W3007542623 @default.
- W3037274327 hasRelatedWork W3015764181 @default.
- W3037274327 hasRelatedWork W3035856201 @default.
- W3037274327 hasRelatedWork W3046768359 @default.
- W3037274327 hasRelatedWork W3095253297 @default.
- W3037274327 hasRelatedWork W3107023854 @default.
- W3037274327 hasRelatedWork W3116635928 @default.
- W3037274327 hasRelatedWork W31519132 @default.
- W3037274327 hasRelatedWork W3155454807 @default.
- W3037274327 hasRelatedWork W3178212073 @default.
- W3037274327 hasRelatedWork W3195860531 @default.
- W3037274327 hasRelatedWork W3211119152 @default.
- W3037274327 hasRelatedWork W35472915 @default.
- W3037274327 isParatext "false" @default.
- W3037274327 isRetracted "false" @default.
- W3037274327 magId "3037274327" @default.
- W3037274327 workType "article" @default.