- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper starts from an important observation: good propensity-score prediction does not necessarily imply good covariate balance in the original covariate space after representation learning. In particular, if Z = \psi(X) is a learned representation, the usual balancing property applies to Z, T \perp Z | e(Z), but this does not automatically imply T \perp X | e(Z). The authors therefore introduce a differentiable propensity-score subclassification module that learns a representation and propensity score; sorts samples by the estimated propensity score; partitions them into contiguous propensity-score subclasses; alternates between neural-network optimization and subclassification optimization. The paper then plugs the module into Dragonnet and BCAUSS, and evaluates it on IHDP and ACIC. The reported results show substantial ATE-bias improvements, especially for Dragonnet + subclassification and BCAUSS + subclassification. The paper also provides an asymptotic equivalence theorem between the hard and smoothed subclassification losses.
Critical strengths
- The motivation is fundamentally correct. The paper identifies an important problem that is sometimes overlooked in neural causal representation learning. A highly predictive propensity model is not necessarily a good balancing representation. In fact, a representation could preserve information useful for predicting treatment while discarding information that is less useful for treatment classification but still important for adjustment.
- Local balance is more informative than only global balance. This is another strong aspect. The paper argues that a global balance metric can hide substantial imbalance in particular propensity-score regions. This is a legitimate concern. The dynamic-programming formulation is also elegant because propensity-score subclasses are constrained to be contiguous after sorting by propensity score.
- The method is genuinely plug-and-play. I like the fact that the authors do not introduce an entirely new causal architecture. The proposed module can be attached to existing representation-learning models equipped with propensity branches.
Critical weaknesses
- The method optimizes only low-order moments, but the paper discusses conditional independence. This is the most important theoretical issue. The optimization objective only measures differences in means and coordinate-wise variances.The balance loss is based on an aggregated SMD of these quantities.
- The representation-learning interpretation is not fully justified. The authors claim that optimizing balance in raw X-space encourages the representation to preserve confounding-relevant information. Therefore the model may be encouraged simply to produce propensity scores that partition observations in a way that makes raw covariates balanced. That is not necessarily equivalent to learning a better low-dimensional representation.
- Dynamic programming is elegant but computationally expensive. This can become expensive for modern causal representation-learning datasets. The authors themselves identify computational cost as a limitation and propose replacing DP with heuristic search in future work. It is difficult to judge whether the proposed method is practically scalable.
Questions for authors
- What exactly is the theoretical target of the method?
- Why should zero SMD imply useful causal balance?
- Can the learned propensity score become intentionally distorted?
- Why are all subclasses equally weighted?