Causal Triplet

홈페이지 · 논문 2편

Causal Triplet is a causal representation learning benchmark featuring not only visually more complex scenes, but also two crucial desiderata commonly overlooked in previous works: 1) An actionable counterfactual setting, where only certain object-level variables allow forcounterfactual observations whereas others do not. 2) An interventional downstream task with an emphasis on out-of-distribution robustness from the independent causal mechanisms principle. Source: Causal Triplet: An Open Challenge for Intervention-centric Causal Representation Learning Image Source: https://arxiv.org/pdf/2301.05169v1.pdf