●builderYou can apply the DAG approximation algorithm to add causally-informed feature attribution to production ML explainability pipelines without prohibitive compute cost.
●researcherThe polynomial-time exact computation result for tree-structured causal graphs is a meaningful theoretical advance over SHAP's worst-case complexity.