LGMENov 17, 2023

Stable Differentiable Causal Discovery

arXiv:2311.10263v234 citationsh-index: 9Has Code
Originality Incremental advance
AI Analysis

This addresses the challenge of inferring causal relationships as directed acyclic graphs for researchers and practitioners in causal inference, representing an incremental improvement over existing differentiable methods.

The paper tackles the problem of numerically unstable differentiable causal discovery methods by proposing Stable Differentiable Causal Discovery (SDCD), which uses a more stable acyclicity constraint and a training procedure for sparse graphs, resulting in improved convergence speed, accuracy, and scalability to thousands of variables.

Inferring causal relationships as directed acyclic graphs (DAGs) is an important but challenging problem. Differentiable Causal Discovery (DCD) is a promising approach to this problem, framing the search as a continuous optimization. But existing DCD methods are numerically unstable, with poor performance beyond tens of variables. In this paper, we propose Stable Differentiable Causal Discovery (SDCD), a new method that improves previous DCD methods in two ways: (1) It employs an alternative constraint for acyclicity; this constraint is more stable, both theoretically and empirically, and fast to compute. (2) It uses a training procedure tailored for sparse causal graphs, which are common in real-world scenarios. We first derive SDCD and prove its stability and correctness. We then evaluate it with both observational and interventional data and on both small-scale and large-scale settings. We find that SDCD outperforms existing methods in both convergence speed and accuracy and can scale to thousands of variables. We provide code at https://github.com/azizilab/sdcd.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes