DAG-SHAP: Feature Attribution in DAG based on Edge Intervention
Keywords: feature attribution, Shapley value, causal intervention
Abstract: Shapley value-based feature attribution methods face challenges in scenarios with complex feature interactions and causal relationships, even when a causal structure is provided. The assumption on the attribution objects of existing methods often deviates from practical scenarios as they cannot capture the exogenous influence of features through each edge in the causal graph, leading to unreasonable interpretations. To overcome these limitations, we propose a novel feature attribution method called DAG-SHAP, which is based on edge intervention. DAG-SHAP treats the exogenous contributions in each ongoing feature edge as an individual attribution object ensuring that both externality and exogenous contributions of features are appropriately captured. Additionally, we introduce an approximation method for efficiently computing DAG-SHAP. Extensive experiments on both synthetic and real datasets validate the effectiveness of DAG-SHAP. Our code can be found in the anonymous repository at \url{https://anonymous.4open.science/r/dag-30F2}.
Primary Area: interpretability and explainable AI
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 7523
Loading