Differentiable Multi-Target Causal Bayesian Experimental Design
This addresses the challenge of efficient causal discovery from finite data where interventions are costly or risky, representing an incremental improvement over prior greedy and black-box methods.
The paper tackles the problem of batch Bayesian optimal experimental design for learning causal models by introducing a gradient-based approach to optimize over multiple intervention target-state pairs, demonstrating that it outperforms existing methods on synthetic datasets.
We introduce a gradient-based approach for the problem of Bayesian optimal experimental design to learn causal models in a batch setting -- a critical component for causal discovery from finite data where interventions can be costly or risky. Existing methods rely on greedy approximations to construct a batch of experiments while using black-box methods to optimize over a single target-state pair to intervene with. In this work, we completely dispose of the black-box optimization techniques and greedy heuristics and instead propose a conceptually simple end-to-end gradient-based optimization procedure to acquire a set of optimal intervention target-state pairs. Such a procedure enables parameterization of the design space to efficiently optimize over a batch of multi-target-state interventions, a setting which has hitherto not been explored due to its complexity. We demonstrate that our proposed method outperforms baselines and existing acquisition strategies in both single-target and multi-target settings across a number of synthetic datasets.