Differentiable Cyclic Causal Discovery Under Unmeasured Confounders

2025-08-11

AI Summary

Real-world causal systems often involve unobserved confounders and feedback loops, posing a fundamental challenge to conventional causal discovery methods that rely on strong assumptions of full observability and acyclicity. To address this, we propose DCCD-CONF—a novel framework for end-to-end differentiable learning of nonlinear, latent-variable–augmented, and cyclic causal graphs. Our approach jointly optimizes the graph structure and the latent confounder distribution via differentiable masking, neural networks for flexible nonlinear mechanism modeling, and an alternating maximum-likelihood optimization strategy. We provide theoretical guarantees on consistency under mild regularity conditions. Extensive experiments on synthetic benchmarks and real-world gene perturbation datasets demonstrate that DCCD-CONF achieves state-of-the-art performance in both causal graph recovery and latent confounder identification.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Understanding causal relationships between variables is fundamental across scientific disciplines. Most causal discovery algorithms rely on two key assumptions: (i) all variables are observed, and (ii) the underlying causal graph is acyclic. While these assumptions simplify theoretical analysis, they are often violated in real-world systems, such as biological networks. Existing methods that account for confounders either assume linearity or struggle with scalability. To address these limitations, we propose DCCD-CONF, a novel framework for differentiable learning of nonlinear cyclic causal graphs in the presence of unmeasured confounders using interventional data. Our approach alternates between optimizing the graph structure and estimating the confounder distribution by maximizing the log-likelihood of the data. Through experiments on synthetic data and real-world gene perturbation datasets, we show that DCCD-CONF outperforms state-of-the-art methods in both causal graph recovery and confounder identification. Additionally, we also provide consistency guarantees for our framework, reinforcing its theoretical soundness.

Problem

• Learning cyclic causal graphs with unmeasured confounders
• Handling nonlinear relationships in causal discovery
• Improving scalability and accuracy in confounder identification

Innovation

• Differentiable learning of cyclic causal graphs
• Handles unmeasured confounders with interventional data
• Optimizes graph structure and confounder distribution