NeFut Logo NeFut
Admin Login

[CS.AI] CAS: A Causal Attribution Score for Explainable AI

Published at: 2026-08-14 22:00 Last updated: 2026-08-16 07:03
#Artificial Intelligence #Explainable AI #Causal Attribution Score

A recent paper on arXiv, titled 'CAS: A Causal Attribution Score for Explainable AI', introduces the Causal Attribution Score (CAS), a compact score architecture for causal explanation. CAS begins with an identified interventional coalition game, allocates the joint intervention contrast with causal Shapley contributions, and converts those raw outcome-scale effects into Local CAS, Signed Local CAS, and two complementary Global CAS summaries. The innovation of CAS is not a new Shapley formula, but a local-to-global causal reporting layer with an explicit intervention target. In the known-truth benchmark, eight repeated primary-interaction simulations (n = 2,200 each, three actions) gave mean Local CAS MAE of 0.107 for coalition-aware CAS, compared with 0.173 for one-at-a-time normalisation and 0.213 for a global normalised absolute ATE vector. Blogger's Review: The Causal Attribution Score CAS provides a compact scoring system for explainable AI, effectively distinguishing between predicting outcomes and explaining heterogeneity in estimated causal effects, allowing us to better understand the model's decision-making process and causal relationships.

Original Source: https://arxiv.org/abs/2608.12555

[h] Back to Home