MissNODAG: Differentiable Cyclic Causal Graph Learning from Incomplete Data
This addresses causal discovery for domains like biology where data is often incomplete and cyclic, offering a method to handle missing not at random data, though it appears incremental as it builds on existing models with extensions for missingness.
The paper tackles the problem of causal discovery in real-world systems with feedback loops and incomplete data by proposing MissNODAG, a differentiable framework that learns cyclic causal graphs and missingness mechanisms from partially observed data, showing effectiveness in synthetic experiments and real-world gene perturbation data.
Causal discovery in real-world systems, such as biological networks, is often complicated by feedback loops and incomplete data. Standard algorithms, which assume acyclic structures or fully observed data, struggle with these challenges. To address this gap, we propose MissNODAG, a differentiable framework for learning both the underlying cyclic causal graph and the missingness mechanism from partially observed data, including data missing not at random. Our framework integrates an additive noise model with an expectation-maximization procedure, alternating between imputing missing values and optimizing the observed data likelihood, to uncover both the cyclic structures and the missingness mechanism. We demonstrate the effectiveness of MissNODAG through synthetic experiments and an application to real-world gene perturbation data.