Differentiable Causal Discovery from Interventional Data

About

Learning a causal directed acyclic graph from data is a challenging task that involves solving a combinatorial problem for which the solution is not always identifiable. A new line of work reformulates this problem as a continuous constrained optimization one, which is solved via the augmented Lagrangian method. However, most methods based on this idea do not make use of interventional data, which can significantly alleviate identifiability issues. This work constitutes a new step in this direction by proposing a theoretically-grounded method based on neural networks that can leverage interventional data. We illustrate the flexibility of the continuous-constrained framework by taking advantage of expressive neural architectures such as normalizing flows. We show that our approach compares favorably to the state of the art in a variety of settings, including perfect and imperfect interventions for which the targeted nodes may even be unknown.

Philippe Brouillard, Sébastien Lachapelle, Alexandre Lacoste, Simon Lacoste-Julien, Alexandre Drouin · 2020

Related benchmarks

Task	Dataset	Result
Causal Discovery	Synthetic Data	Runtime837.8	57
Causal Structure Learning	Sachs	SHD21	38
Causal Discovery	Semantic Causal Environment observation-only	F1 Score33.8	15
Graph Recovery	Synthetic Nonlinear SEM Gaussian Noise (test)	AUPRC20.4	15
Graph Recovery	Synthetic Nonlinear SEM Exponential Noise (test)	AUPRC15.8	15
Graph Recovery	Synthetic Nonlinear SEM (Gumbel Noise) (test)	AUPRC0.109	15
Causal Discovery	Synthetic Data Observation-only (1000 samples)	Rank13	15
Causal Discovery	K562 Perturb-seq CausalBench	W Score18.3	13
Causal Discovery	RPE1 Perturb-seq CausalBench	W Score19.4	13
Causal Discovery	SACHS p = 11, s = 20, n = 100 (real flow cytometry)	F1 Score22	13