Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313447757> ?p ?o ?g. }
Showing items 1 to 67 of
67
with 100 items per page.
- W4313447757 abstract "Explainable AI transforms opaque decision strategies of ML models into explanations that are interpretable by the user, for example, identifying the contribution of each input feature to the prediction at hand. Such explanations, however, entangle the potentially multiple factors that enter into the overall complex decision strategy. We propose to disentangle explanations by finding relevant subspaces in activation space that can be mapped to more abstract human-understandable concepts and enable a joint attribution on concepts and input features. To automatically extract the desired representation, we propose new subspace analysis formulations that extend the principle of PCA and subspace analysis to explanations. These novel analyses, which we call principal relevant component analysis (PRCA) and disentangled relevant subspace analysis (DRSA), optimize relevance of projected activations rather than the more traditional variance or kurtosis. This enables a much stronger focus on subspaces that are truly relevant for the prediction and the explanation, in particular, ignoring activations or concepts to which the prediction model is invariant. Our approach is general enough to work alongside common attribution techniques such as Shapley Value, Integrated Gradients, or LRP. Our proposed methods show to be practically useful and compare favorably to the state of the art as demonstrated on benchmarks and three use cases." @default.
- W4313447757 created "2023-01-06" @default.
- W4313447757 creator A5011367405 @default.
- W4313447757 creator A5026240645 @default.
- W4313447757 creator A5026857798 @default.
- W4313447757 creator A5072994165 @default.
- W4313447757 date "2022-12-30" @default.
- W4313447757 modified "2023-10-18" @default.
- W4313447757 title "Disentangled Explanations of Neural Network Predictions by Finding Relevant Subspaces" @default.
- W4313447757 doi "https://doi.org/10.48550/arxiv.2212.14855" @default.
- W4313447757 hasPublicationYear "2022" @default.
- W4313447757 type Work @default.
- W4313447757 citedByCount "0" @default.
- W4313447757 crossrefType "posted-content" @default.
- W4313447757 hasAuthorship W4313447757A5011367405 @default.
- W4313447757 hasAuthorship W4313447757A5026240645 @default.
- W4313447757 hasAuthorship W4313447757A5026857798 @default.
- W4313447757 hasAuthorship W4313447757A5072994165 @default.
- W4313447757 hasBestOaLocation W43134477571 @default.
- W4313447757 hasConcept C119857082 @default.
- W4313447757 hasConcept C12362212 @default.
- W4313447757 hasConcept C153180895 @default.
- W4313447757 hasConcept C154945302 @default.
- W4313447757 hasConcept C158154518 @default.
- W4313447757 hasConcept C17744445 @default.
- W4313447757 hasConcept C190470478 @default.
- W4313447757 hasConcept C199539241 @default.
- W4313447757 hasConcept C2524010 @default.
- W4313447757 hasConcept C27438332 @default.
- W4313447757 hasConcept C2776359362 @default.
- W4313447757 hasConcept C32834561 @default.
- W4313447757 hasConcept C33923547 @default.
- W4313447757 hasConcept C37914503 @default.
- W4313447757 hasConcept C41008148 @default.
- W4313447757 hasConcept C94625758 @default.
- W4313447757 hasConceptScore W4313447757C119857082 @default.
- W4313447757 hasConceptScore W4313447757C12362212 @default.
- W4313447757 hasConceptScore W4313447757C153180895 @default.
- W4313447757 hasConceptScore W4313447757C154945302 @default.
- W4313447757 hasConceptScore W4313447757C158154518 @default.
- W4313447757 hasConceptScore W4313447757C17744445 @default.
- W4313447757 hasConceptScore W4313447757C190470478 @default.
- W4313447757 hasConceptScore W4313447757C199539241 @default.
- W4313447757 hasConceptScore W4313447757C2524010 @default.
- W4313447757 hasConceptScore W4313447757C27438332 @default.
- W4313447757 hasConceptScore W4313447757C2776359362 @default.
- W4313447757 hasConceptScore W4313447757C32834561 @default.
- W4313447757 hasConceptScore W4313447757C33923547 @default.
- W4313447757 hasConceptScore W4313447757C37914503 @default.
- W4313447757 hasConceptScore W4313447757C41008148 @default.
- W4313447757 hasConceptScore W4313447757C94625758 @default.
- W4313447757 hasLocation W43134477571 @default.
- W4313447757 hasOpenAccess W4313447757 @default.
- W4313447757 hasPrimaryLocation W43134477571 @default.
- W4313447757 hasRelatedWork W1996238891 @default.
- W4313447757 hasRelatedWork W202122401 @default.
- W4313447757 hasRelatedWork W2065962751 @default.
- W4313447757 hasRelatedWork W2108334821 @default.
- W4313447757 hasRelatedWork W2117190566 @default.
- W4313447757 hasRelatedWork W2121787161 @default.
- W4313447757 hasRelatedWork W2126175004 @default.
- W4313447757 hasRelatedWork W2128580395 @default.
- W4313447757 hasRelatedWork W2170762631 @default.
- W4313447757 hasRelatedWork W2185562971 @default.
- W4313447757 isParatext "false" @default.
- W4313447757 isRetracted "false" @default.
- W4313447757 workType "article" @default.