SCOUT: Cyclic Causal Discovery Under Soft Interventions with Unknown Targets

Learning causal relationships between variables from data is a fundamental research area with many applications across disciplines. Most of the existing causal discovery algorithms rely on the assumptions that (i) the underlying system is acyclic, (ii) the exogenous noise variables are Gaussian, and (iii) that the intervention targets for the data generating experiments are known. While these assumptions simplify the analysis, they are violated in real-life systems. Most existing methods that address these issues either assume the underlying model is linear or are constrained to operate in limited interventional settings. To that end, we propose SCOUT, a novel causal discovery framework to learn nonlinear causal cyclic relationships from soft interventional data with unknown targets. Our main approach maximizes the data log-likelihood to recover the graph structure, using two normalizing-flow architectures—contractive residual flows and neural spline flows. By conducting experiments on synthetic and real-world data, we show that SCOUT outperforms state-of-the-art methods in both causal graph and unknown target recovery across various interventional and noise settings.

Many scientific systems, such as gene regulation and other biological networks, contain feedback: one factor can influence another, which can in turn influence the first. Learning these cause-and-effect relationships from data is difficult, especially when experiments only partially change the system and researchers do not know exactly which factors were changed. We introduce SCOUT, a machine learning method designed for this more realistic setting. SCOUT looks across data from several experiments and jointly learns both the hidden map of cause-and-effect relationships and which variables were affected in each experiment. It can also handle complex behavior and noise patterns that are common in real data. In tests on simulated systems and gene-activity data, SCOUT recovered causal relationships and experimental targets more accurately than existing methods and gave better predictions for unseen experimental conditions. This work can help scientists better understand complex systems and plan future experiments more effectively.