Colin Rowat — accepted AI-conference papers

2022

Joint Shapley values: a measure of joint feature importance

Chris Harris, Richard Pymar, Colin Rowat

ICLR 2022poster

The Shapley value is one of the most widely used measures of feature importance partly as it measures a feature's average effect on a model's prediction. We introduce joint Shapley values, which directly extend Shapley's axioms and intuitions: joint Shapley values measure a set of features' average…

Cited by 34 Source ↗PDF ↗code ↗Scholar ↗

2020

Asymmetric Shapley values: incorporating causal knowledge into model-agnostic explainability

Christopher Frye, Colin Rowat, Ilya Feige

NeurIPS 2020poster

Explaining AI systems is fundamental both to the development of high performing models and to the trust placed in them by their users. The Shapley framework for explainability has strength in its general applicability combined with its precise, rigorous foundation: it provides a common, model-agnost…

Cited by 271 Source ↗PDF ↗Scholar ↗