Answer-set programs for reasoning about counterfactual interventions and responsibility scores for classification

From MaRDI portal
Publication:6166547




Abstract: We describe how answer-set programs can be used to declaratively specify counterfactual interventions on entities under classification, and reason about them. In particular, they can be used to define and compute responsibility scores as attribution-based explanations for outcomes from classification models. The approach allows for the inclusion of domain knowledge and supports query answering. A detailed example with a naive-Bayes classifier is presented.









This page was built for publication: Answer-set programs for reasoning about counterfactual interventions and responsibility scores for classification

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6166547)