Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287162490> ?p ?o ?g. }
Showing items 1 to 65 of
65
with 100 items per page.
- W4287162490 abstract "Explaining deep learning model inferences is a promising venue for scientific understanding, improving safety, uncovering hidden biases, evaluating fairness, and beyond, as argued by many scholars. One of the principal benefits of counterfactual explanations is allowing users to explore what-if scenarios through what does not and cannot exist in the data, a quality that many other forms of explanation such as heatmaps and influence functions are inherently incapable of doing. However, most previous work on generative explainability cannot disentangle important concepts effectively, produces unrealistic examples, or fails to retain relevant information. We propose a novel approach, DISSECT, that jointly trains a generator, a discriminator, and a concept disentangler to overcome such challenges using little supervision. DISSECT generates Concept Traversals (CTs), defined as a sequence of generated examples with increasing degrees of concepts that influence a classifier's decision. By training a generative model from a classifier's signal, DISSECT offers a way to discover a classifier's inherent notion of distinct concepts automatically rather than rely on user-predefined concepts. We show that DISSECT produces CTs that (1) disentangle several concepts, (2) are influential to a classifier's decision and are coupled to its reasoning due to joint training (3), are realistic, (4) preserve relevant information, and (5) are stable across similar inputs. We validate DISSECT on several challenging synthetic and realistic datasets where previous methods fall short of satisfying desirable criteria for interpretability and show that it performs consistently well and better than existing methods. Finally, we present experiments showing applications of DISSECT for detecting potential biases of a classifier and identifying spurious artifacts that impact predictions." @default.
- W4287162490 created "2022-07-25" @default.
- W4287162490 creator A5006517585 @default.
- W4287162490 creator A5022368228 @default.
- W4287162490 creator A5041405652 @default.
- W4287162490 creator A5042346519 @default.
- W4287162490 creator A5043714597 @default.
- W4287162490 creator A5087366916 @default.
- W4287162490 date "2021-05-31" @default.
- W4287162490 modified "2023-09-26" @default.
- W4287162490 title "DISSECT: Disentangled Simultaneous Explanations via Concept Traversals" @default.
- W4287162490 doi "https://doi.org/10.48550/arxiv.2105.15164" @default.
- W4287162490 hasPublicationYear "2021" @default.
- W4287162490 type Work @default.
- W4287162490 citedByCount "0" @default.
- W4287162490 crossrefType "posted-content" @default.
- W4287162490 hasAuthorship W4287162490A5006517585 @default.
- W4287162490 hasAuthorship W4287162490A5022368228 @default.
- W4287162490 hasAuthorship W4287162490A5041405652 @default.
- W4287162490 hasAuthorship W4287162490A5042346519 @default.
- W4287162490 hasAuthorship W4287162490A5043714597 @default.
- W4287162490 hasAuthorship W4287162490A5087366916 @default.
- W4287162490 hasBestOaLocation W42871624901 @default.
- W4287162490 hasConcept C108650721 @default.
- W4287162490 hasConcept C119857082 @default.
- W4287162490 hasConcept C154945302 @default.
- W4287162490 hasConcept C15744967 @default.
- W4287162490 hasConcept C167966045 @default.
- W4287162490 hasConcept C2779803651 @default.
- W4287162490 hasConcept C2781067378 @default.
- W4287162490 hasConcept C39890363 @default.
- W4287162490 hasConcept C41008148 @default.
- W4287162490 hasConcept C76155785 @default.
- W4287162490 hasConcept C77805123 @default.
- W4287162490 hasConcept C94915269 @default.
- W4287162490 hasConcept C95623464 @default.
- W4287162490 hasConceptScore W4287162490C108650721 @default.
- W4287162490 hasConceptScore W4287162490C119857082 @default.
- W4287162490 hasConceptScore W4287162490C154945302 @default.
- W4287162490 hasConceptScore W4287162490C15744967 @default.
- W4287162490 hasConceptScore W4287162490C167966045 @default.
- W4287162490 hasConceptScore W4287162490C2779803651 @default.
- W4287162490 hasConceptScore W4287162490C2781067378 @default.
- W4287162490 hasConceptScore W4287162490C39890363 @default.
- W4287162490 hasConceptScore W4287162490C41008148 @default.
- W4287162490 hasConceptScore W4287162490C76155785 @default.
- W4287162490 hasConceptScore W4287162490C77805123 @default.
- W4287162490 hasConceptScore W4287162490C94915269 @default.
- W4287162490 hasConceptScore W4287162490C95623464 @default.
- W4287162490 hasLocation W42871624901 @default.
- W4287162490 hasOpenAccess W4287162490 @default.
- W4287162490 hasPrimaryLocation W42871624901 @default.
- W4287162490 hasRelatedWork W2562221994 @default.
- W4287162490 hasRelatedWork W2950863313 @default.
- W4287162490 hasRelatedWork W2977603687 @default.
- W4287162490 hasRelatedWork W2979484800 @default.
- W4287162490 hasRelatedWork W3125159465 @default.
- W4287162490 hasRelatedWork W4213330993 @default.
- W4287162490 hasRelatedWork W4282813524 @default.
- W4287162490 hasRelatedWork W4287162490 @default.
- W4287162490 hasRelatedWork W4287356576 @default.
- W4287162490 hasRelatedWork W4287370216 @default.
- W4287162490 isParatext "false" @default.
- W4287162490 isRetracted "false" @default.
- W4287162490 workType "article" @default.