On Inductive Biases for Heterogeneous Treatment Effect Estimation
This work addresses the challenge of accurately estimating treatment effects in causal inference, particularly when treatment effects may be absent, which is important for fields like medicine and policy-making, but it is incremental as it builds on existing methods by comparing and enhancing them.
The paper tackled the problem of improving conditional average treatment effect estimation by exploiting structural similarities in potential outcomes under different treatments, and found that three proposed learning strategies based on regularization, reparametrization, and multi-task architectures led to substantial improvements over baselines in semi-synthetic experiments.
We investigate how to exploit structural similarities of an individual's potential outcomes (POs) under different treatments to obtain better estimates of conditional average treatment effects in finite samples. Especially when it is unknown whether a treatment has an effect at all, it is natural to hypothesize that the POs are similar - yet, some existing strategies for treatment effect estimation employ regularization schemes that implicitly encourage heterogeneity even when it does not exist and fail to fully make use of shared structure. In this paper, we investigate and compare three end-to-end learning strategies to overcome this problem - based on regularization, reparametrization and a flexible multi-task architecture - each encoding inductive bias favoring shared behavior across POs. To build understanding of their relative strengths, we implement all strategies using neural networks and conduct a wide range of semi-synthetic experiments. We observe that all three approaches can lead to substantial improvements upon numerous baselines and gain insight into performance differences across various experimental settings.