TabCausal: Pretraining Across Causal Environments for Tabular Causal Discovery

2026-05-29

AI Summary

This work addresses the challenge of efficiently and reliably recovering transferable causal graph structures from both observational and interventional data. To this end, the authors propose a data-driven foundational model for causal discovery that, through pretraining across diverse causal environments, directly maps tabular data to causal graphs—bypassing the need for dataset-specific search or optimization inherent in traditional approaches. The method innovatively incorporates a dynamic task construction strategy that integrates multiple graph priors, causal mechanisms, noise models, and intervention types. A large-scale benchmark comprising synthetic and semantically grounded structural causal models (SCMs) is introduced, enhanced with LLM-assisted semantic validation. Experiments demonstrate that the proposed approach consistently outperforms existing methods on both synthetic and real-world semantic benchmarks, exhibiting superior causal structure recovery and out-of-distribution generalization, particularly when interventional data are available.

Abstract

Causal discovery aims to recover directed causal relations from observational and interventional data, providing a basis for mechanistic understanding and reliable decision-making. Causal discovery foundation models (CDFMs) seek to amortize this problem by mapping a dataset directly to a causal graph in a single forward pass, avoiding per-dataset testing, search, or optimization. However, existing CDFMs remain limited, often failing to consistently match strong classical methods, and we find that a key bottleneck is how causal pretraining tasks are constructed. Based on this observation, we propose TabCausal, a data-driven CDFM trained with broad causal pretraining over diverse graph priors, structural mechanisms, noise models, dimensions, sample sizes, and intervention regimes. A dynamic task construction strategy composes these causal environments into varied discovery tasks, enabling more transferable structural learning from observational and mixed-interventional data. On large-scale synthetic benchmarks, TabCausal achieves better macro-averaged performance than a diverse set of causal discovery baselines. To further bridge abstract synthetic generators and realistic causal reasoning scenarios, we introduce a protocol-guided and LLM-audited semantic causal environment benchmark, where domain-grounded SCMs generate interpretable observational and interventional datasets for out-of-distribution analysis. Across both synthetic and semantic environments, TabCausal demonstrates robust structure recovery, especially under interventional evidence, highlighting broad causal pretraining as a key ingredient for transferable amortized causal discovery.

Problem

• causal discovery
• foundation models
• pretraining
• causal graphs
• interventional data

Innovation

• causal discovery foundation model
• broad causal pretraining
• dynamic task construction
• semantic causal benchmark
• amortized causal inference