The Challenge of Imputation in Explainable Artificial Intelligence Models
This addresses a safety issue in explainable AI for users relying on transparent models, but it appears incremental as it builds on known concerns about imputation.
The paper tackles the problem that imputing missing values in datasets can lead to counterfactual scenarios, potentially causing unsafe outcomes when acting on explanations from AI models, and it explores problematic settings and proposes ways to address them.
Explainable models in Artificial Intelligence are often employed to ensure transparency and accountability of AI systems. The fidelity of the explanations are dependent upon the algorithms used as well as on the fidelity of the data. Many real world datasets have missing values that can greatly influence explanation fidelity. The standard way to deal with such scenarios is imputation. This can, however, lead to situations where the imputed values may correspond to a setting which refer to counterfactuals. Acting on explanations from AI models with imputed values may lead to unsafe outcomes. In this paper, we explore different settings where AI models with imputation can be problematic and describe ways to address such scenarios.