Dagma-DCE: Interpretable, Non-Parametric Differentiable Causal Discovery
This work addresses interpretability issues in causal discovery for researchers, though it appears incremental as it builds on existing differentiable methods.
The paper tackles the problem of opaque proxies for independence in differentiable causal discovery by introducing Dagma-DCE, which uses an interpretable measure of causal strength, achieving state-of-the-art performance in simulated datasets.
We introduce Dagma-DCE, an interpretable and model-agnostic scheme for differentiable causal discovery. Current non- or over-parametric methods in differentiable causal discovery use opaque proxies of ``independence'' to justify the inclusion or exclusion of a causal relationship. We show theoretically and empirically that these proxies may be arbitrarily different than the actual causal strength. Juxtaposed to existing differentiable causal discovery algorithms, \textsc{Dagma-DCE} uses an interpretable measure of causal strength to define weighted adjacency matrices. In a number of simulated datasets, we show our method achieves state-of-the-art level performance. We additionally show that \textsc{Dagma-DCE} allows for principled thresholding and sparsity penalties by domain-experts. The code for our method is available open-source at https://github.com/DanWaxman/DAGMA-DCE, and can easily be adapted to arbitrary differentiable models.