Cross-Domain Transfer Performance of Tabular Foundation Models Pretrained on Synthetic Adversarial Data Versus Real-World Datasets

The development of tabular foundation models (TFMs) has accelerated in recent years, showing strong potential to outperform traditional ML methods for structured data. A key finding is that TFMs can be pretrained entirely on synthetic datasets, opening opportunities to design data generators that encourage desirable model properties. Prior work has mainly focused on crafting high-quality priors over generators to improve overall pretraining performance. Our insight is that parameterizing the generator distribution enables an adversarial robustness perspective: during training, we can adapt the

Research goal: How do tabular foundation models pretrained on synthetic data with adversarial noise perform in cross-domain transfer learning tasks compared to models pretrained on real-world datasets, as measured by accuracy and robustness on TabMNAR and TabCI benchmarks?

Autonomous synthesis report generated by SOVEREIGN Research Kernel. Tribunal consensus score: 7.8/10.