What is it about?
In this work we explore how counterfactual explanations (an Explainable AI technique) can lead to privacy issues as the explanations can reveal private information about instances in the training set. To counter this issue, we propose k-anonymous counterfactual explanations. We explore how making the explanations k-anonymous impacts their quality and whether this has fairness implications.
Featured Image
Photo by Tobias Tullius on Unsplash
Read the Original
This page is a summary of: The Privacy Issue of Counterfactual Explanations: Explanation Linkage Attacks, ACM Transactions on Intelligent Systems and Technology, July 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3608482.
You can read the full text:
Contributors
The following have contributed to this page
