AI Paper Atlas

Research map / Probabilistic and scientific modeling / Causal discovery

The Fairness Hierarchy: A viewpoint from causal inference

ChengBo Zhang, Zhen Yao, Hao Pang, Changcheng Li

ICML 2026 · Probabilistic and scientific modeling › Causal discovery · approach: Causal approaches to fairness

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
intervene on A only (single-sensitive) instead of also on confounder Z (ablation)path-specific counterfactual prediction · Synthetic SCMP(Y_hat=1 | A=0) vs reference group: 0.528 vs 0.663 reference groupHurt
add distributional intervention on confounder Z (multi-sensitive Y(2))path-specific counterfactual prediction · Synthetic SCMgroup positive prediction rate gap: 0.270 -> 0.663 for A=0 (reference 0.663)Improved
impose stronger fairness constraint (Y(0) to Y(2))path-specific counterfactual prediction · Synthetic SCMoverall accuracy: 0.923 -> 0.757Hurt
allow admissible paths via Z->A->M->Y (Y(3) instead of Y(2)) (ablation)multi-sensitive intervention · Synthetic SCMoverall accuracy: 0.757 -> 0.838Improved
misspecify working SCM with extra edge Z->W (ablation)Y(2) intervention · Synthetic SCMgroup positive prediction probability: 0.654 vs 0.663No clear effect
remove confounder Z from the working causal graph (ablation)Y(2) intervention · Synthetic SCMoverall accuracy / group disparity: accuracy 0.683; positive rates 0.762 vs 0.663Hurt
Kusner-style residual predictor instead of multi-sensitive interventionfair prediction · Synthetic SCMgroup positive prediction probability: 0.379 for A=0 and 0.600 for A=1Hurt
block direct path A->Y while keeping A->M->YCOMPAS predictor · COMPASaccuracy: 0.657 -> 0.656No clear effect
further restrict Z to flow only through Z->A->M->YCOMPAS predictor · COMPASaccuracy: 0.656 -> 0.618Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy