Improving TabPFN’s Synthetic Data Generation by Integrating Causal Structure

Davide Tugnoli, Andrea De Lorenzo, Marco Virgolin, Giovanni Cinà
Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, PMLR 337:6807-6850, 2026.

Abstract

Synthetic tabular data generation addresses data scarcity and privacy constraints in a variety of domains. Tabular Prior-Data Fitted Network ({TabPFN}), a recent foundation model for tabular data, has been shown capable of generating high-quality synthetic tabular data. However, {TabPFN} is autoregressive: features are generated sequentially by conditioning on the previous ones, depending on the order in which they appear in the input data. We demonstrate that when the feature order conflicts with causal structure, the model produces spurious correlations that impair its ability to generate synthetic data and preserve causal effects. We address this limitation by integrating causal structure into {TabPFN}’s generation process through two complementary approaches: Directed Acyclic Graph ({DAG})-aware conditioning, which samples each variable given its causal parents, and a partially directed acyclic graph (PDAG)-based strategy for scenarios with partial causal knowledge. We evaluate these approaches on controlled benchmarks and six CSuite datasets, assessing structural fidelity, distributional quality, and Average Treatment Effect ({ATE}) preservation. Across most settings, {DAG}-aware conditioning improves the quality and stability of synthetic data relative to vanilla {TabPFN}. Under partial causal knowledge, the oracle partially directed acyclic graph (oracle-PDAG), which orients only the edges into the colliders, shows moderate gains, while the benefit of a Completed Partially Directed Acyclic Graph (CPDAG) discovered from data depends on how well the causal structure is recovered. These results indicate that reliable causal structure, even partial, can be injected into {TabPFN} at inference time, without parameter updates, to improve synthetic data quality.

Cite this Paper

BibTeX
@InProceedings{pmlr-v337-tugnoli26a, title = {Improving {TabPFN}’s Synthetic Data Generation by Integrating Causal Structure}, author = {Tugnoli, Davide and De Lorenzo, Andrea and Virgolin, Marco and Cin\`{a}, Giovanni}, booktitle = {Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence}, pages = {6807--6850}, year = {2026}, editor = {Perković, Emilija and Malinsky, Daniel}, volume = {337}, series = {Proceedings of Machine Learning Research}, month = {17--21 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v337/main/assets/tugnoli26a/tugnoli26a.pdf}, url = {https://proceedings.mlr.press/v337/tugnoli26a.html}, abstract = {Synthetic tabular data generation addresses data scarcity and privacy constraints in a variety of domains. Tabular Prior-Data Fitted Network ({TabPFN}), a recent foundation model for tabular data, has been shown capable of generating high-quality synthetic tabular data. However, {TabPFN} is autoregressive: features are generated sequentially by conditioning on the previous ones, depending on the order in which they appear in the input data. We demonstrate that when the feature order conflicts with causal structure, the model produces spurious correlations that impair its ability to generate synthetic data and preserve causal effects. We address this limitation by integrating causal structure into {TabPFN}’s generation process through two complementary approaches: Directed Acyclic Graph ({DAG})-aware conditioning, which samples each variable given its causal parents, and a partially directed acyclic graph (PDAG)-based strategy for scenarios with partial causal knowledge. We evaluate these approaches on controlled benchmarks and six CSuite datasets, assessing structural fidelity, distributional quality, and Average Treatment Effect ({ATE}) preservation. Across most settings, {DAG}-aware conditioning improves the quality and stability of synthetic data relative to vanilla {TabPFN}. Under partial causal knowledge, the oracle partially directed acyclic graph (oracle-PDAG), which orients only the edges into the colliders, shows moderate gains, while the benefit of a Completed Partially Directed Acyclic Graph (CPDAG) discovered from data depends on how well the causal structure is recovered. These results indicate that reliable causal structure, even partial, can be injected into {TabPFN} at inference time, without parameter updates, to improve synthetic data quality.} }
Endnote
%0 Conference Paper %T Improving TabPFN’s Synthetic Data Generation by Integrating Causal Structure %A Davide Tugnoli %A Andrea De Lorenzo %A Marco Virgolin %A Giovanni Cinà %B Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2026 %E Emilija Perković %E Daniel Malinsky %F pmlr-v337-tugnoli26a %I PMLR %P 6807--6850 %U https://proceedings.mlr.press/v337/tugnoli26a.html %V 337 %X Synthetic tabular data generation addresses data scarcity and privacy constraints in a variety of domains. Tabular Prior-Data Fitted Network ({TabPFN}), a recent foundation model for tabular data, has been shown capable of generating high-quality synthetic tabular data. However, {TabPFN} is autoregressive: features are generated sequentially by conditioning on the previous ones, depending on the order in which they appear in the input data. We demonstrate that when the feature order conflicts with causal structure, the model produces spurious correlations that impair its ability to generate synthetic data and preserve causal effects. We address this limitation by integrating causal structure into {TabPFN}’s generation process through two complementary approaches: Directed Acyclic Graph ({DAG})-aware conditioning, which samples each variable given its causal parents, and a partially directed acyclic graph (PDAG)-based strategy for scenarios with partial causal knowledge. We evaluate these approaches on controlled benchmarks and six CSuite datasets, assessing structural fidelity, distributional quality, and Average Treatment Effect ({ATE}) preservation. Across most settings, {DAG}-aware conditioning improves the quality and stability of synthetic data relative to vanilla {TabPFN}. Under partial causal knowledge, the oracle partially directed acyclic graph (oracle-PDAG), which orients only the edges into the colliders, shows moderate gains, while the benefit of a Completed Partially Directed Acyclic Graph (CPDAG) discovered from data depends on how well the causal structure is recovered. These results indicate that reliable causal structure, even partial, can be injected into {TabPFN} at inference time, without parameter updates, to improve synthetic data quality.
APA
Tugnoli, D., De Lorenzo, A., Virgolin, M. & Cinà, G.. (2026). Improving TabPFN’s Synthetic Data Generation by Integrating Causal Structure. Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 337:6807-6850 Available from https://proceedings.mlr.press/v337/tugnoli26a.html.

Related Material