Matches in SemOpenAlex for { <https://semopenalex.org/work/W3109440167> ?p ?o ?g. }
- W3109440167 endingPage "11554" @default.
- W3109440167 startingPage "11545" @default.
- W3109440167 abstract "Neural network visualization techniques mark image locations by their relevancy to the network's classification. Existing methods are effective in highlighting the regions that affect the resulting classification the most. However, as we show, these methods are limited in their ability to identify the support for alternative classifications, an effect we name the saliency bias hypothesis. In this work, we integrate two lines of research: gradient-based methods and attribution-based methods, and develop an algorithm that provides per-class explainability. The algorithm back-projects the per pixel local influence, in a manner that is guided by the local attributions, while correcting for salient features that would otherwise bias the explanation. In an extensive battery of experiments, we demonstrate the ability of our methods to class-specific visualization, and not just the predicted label. Remarkably, the method obtains state of the art results in benchmarks that are commonly applied to gradient-based methods as well as in those that are employed mostly for evaluating attribution methods. Using a new unsupervised procedure, our method is also successful in demonstrating that self-supervised methods learn semantic information. Our code is available at: https://github.com/shirgur/AGFVisualization." @default.
- W3109440167 created "2020-12-07" @default.
- W3109440167 creator A5011488127 @default.
- W3109440167 creator A5078102229 @default.
- W3109440167 creator A5080576073 @default.
- W3109440167 date "2021-05-18" @default.
- W3109440167 modified "2023-10-05" @default.
- W3109440167 title "Visualization of Supervised and Self-Supervised Neural Networks via Attribution Guided Factorization" @default.
- W3109440167 cites W1603679155 @default.
- W3109440167 cites W1787224781 @default.
- W3109440167 cites W1849277567 @default.
- W3109440167 cites W2117539524 @default.
- W3109440167 cites W2117741877 @default.
- W3109440167 cites W2190008860 @default.
- W3109440167 cites W2195388612 @default.
- W3109440167 cites W2295107390 @default.
- W3109440167 cites W2336525064 @default.
- W3109440167 cites W2346578521 @default.
- W3109440167 cites W2516884085 @default.
- W3109440167 cites W2594633041 @default.
- W3109440167 cites W2605409611 @default.
- W3109440167 cites W2626639386 @default.
- W3109440167 cites W2776207810 @default.
- W3109440167 cites W2799124825 @default.
- W3109440167 cites W2891612330 @default.
- W3109440167 cites W2903165775 @default.
- W3109440167 cites W2928920860 @default.
- W3109440167 cites W2951264787 @default.
- W3109440167 cites W2962851944 @default.
- W3109440167 cites W2962858109 @default.
- W3109440167 cites W2962862931 @default.
- W3109440167 cites W2962867364 @default.
- W3109440167 cites W2963081790 @default.
- W3109440167 cites W2963715038 @default.
- W3109440167 cites W2969115771 @default.
- W3109440167 cites W2970005358 @default.
- W3109440167 cites W2970852027 @default.
- W3109440167 cites W2972107091 @default.
- W3109440167 cites W2988157455 @default.
- W3109440167 cites W2989731761 @default.
- W3109440167 cites W2995181141 @default.
- W3109440167 cites W3101609372 @default.
- W3109440167 cites W3110446398 @default.
- W3109440167 doi "https://doi.org/10.1609/aaai.v35i13.17374" @default.
- W3109440167 hasPublicationYear "2021" @default.
- W3109440167 type Work @default.
- W3109440167 sameAs 3109440167 @default.
- W3109440167 citedByCount "7" @default.
- W3109440167 countsByYear W31094401672021 @default.
- W3109440167 countsByYear W31094401672022 @default.
- W3109440167 countsByYear W31094401672023 @default.
- W3109440167 crossrefType "journal-article" @default.
- W3109440167 hasAuthorship W3109440167A5011488127 @default.
- W3109440167 hasAuthorship W3109440167A5078102229 @default.
- W3109440167 hasAuthorship W3109440167A5080576073 @default.
- W3109440167 hasBestOaLocation W31094401671 @default.
- W3109440167 hasConcept C119857082 @default.
- W3109440167 hasConcept C124101348 @default.
- W3109440167 hasConcept C143299363 @default.
- W3109440167 hasConcept C153180895 @default.
- W3109440167 hasConcept C154945302 @default.
- W3109440167 hasConcept C15744967 @default.
- W3109440167 hasConcept C160633673 @default.
- W3109440167 hasConcept C177264268 @default.
- W3109440167 hasConcept C199360897 @default.
- W3109440167 hasConcept C2776760102 @default.
- W3109440167 hasConcept C2777212361 @default.
- W3109440167 hasConcept C2780719617 @default.
- W3109440167 hasConcept C36464697 @default.
- W3109440167 hasConcept C41008148 @default.
- W3109440167 hasConcept C50644808 @default.
- W3109440167 hasConcept C77805123 @default.
- W3109440167 hasConceptScore W3109440167C119857082 @default.
- W3109440167 hasConceptScore W3109440167C124101348 @default.
- W3109440167 hasConceptScore W3109440167C143299363 @default.
- W3109440167 hasConceptScore W3109440167C153180895 @default.
- W3109440167 hasConceptScore W3109440167C154945302 @default.
- W3109440167 hasConceptScore W3109440167C15744967 @default.
- W3109440167 hasConceptScore W3109440167C160633673 @default.
- W3109440167 hasConceptScore W3109440167C177264268 @default.
- W3109440167 hasConceptScore W3109440167C199360897 @default.
- W3109440167 hasConceptScore W3109440167C2776760102 @default.
- W3109440167 hasConceptScore W3109440167C2777212361 @default.
- W3109440167 hasConceptScore W3109440167C2780719617 @default.
- W3109440167 hasConceptScore W3109440167C36464697 @default.
- W3109440167 hasConceptScore W3109440167C41008148 @default.
- W3109440167 hasConceptScore W3109440167C50644808 @default.
- W3109440167 hasConceptScore W3109440167C77805123 @default.
- W3109440167 hasIssue "13" @default.
- W3109440167 hasLocation W31094401671 @default.
- W3109440167 hasLocation W31094401672 @default.
- W3109440167 hasOpenAccess W3109440167 @default.
- W3109440167 hasPrimaryLocation W31094401671 @default.
- W3109440167 hasRelatedWork W2114184889 @default.
- W3109440167 hasRelatedWork W2136485282 @default.
- W3109440167 hasRelatedWork W2546871836 @default.
- W3109440167 hasRelatedWork W2726222394 @default.
- W3109440167 hasRelatedWork W2961085424 @default.