Engineering Regression Without Real-Data Training: Domain Adaptation for Tabular Foundation Models Using Multi-Dataset Embeddings

TL;DR

This paper presents a domain adaptation method for engineering regression tasks using synthetic data curation, improving foundation model performance without real data training by identifying engineering-like synthetic datasets.

Contribution

It introduces TREDBench, a collection of real-world datasets, and a synthetic data curation approach that enhances foundation model transfer to engineering domains.

Findings

Synthetic data curation improves predictive accuracy.

Synthetic-only adaptation outperforms baseline models.

Data efficiency gains of 1.75x to 4.44x achieved.

Abstract

Predictive modeling in engineering applications has long been dominated by bespoke models and small, siloed tabular datasets, limiting the applicability of large-scale learning approaches. Despite recent progress in tabular foundation models, the resulting synthetic training distributions used for pre-training may not reflect the statistical structure of engineering data, limiting transfer to engineering regression. We introduce TREDBench, a curated collection of 83 real-world tabular regression datasets with expert engineering/non-engineering labels, and use TabPFN 2.5's dataset-level embedding to study domain structure in a common representation space. We find that engineering datasets are partially distinguishable from non-engineering datasets, while standard procedurally generated datasets are highly distinguishable from engineering datasets, revealing a substantial synthetic-real domain gap. To bridge this gap without training on real engineering samples, we propose an embedding-guided synthetic data curation method: we generate and identify "engineering-like" synthetic datasets, and perform continued pre-training of TabPFN 2.5 using only the selected synthetic tasks. Across 35 engineering regression datasets, this synthetic-only adaptation improves predictive accuracy and data efficiency, outperforming TabPFN 2.5 on 29/35 datasets and AutoGluon on 27/35, with mean multiplicative data-efficiency gains of 1.75x and 4.44x, respectively. More broadly, our results indicate that principled synthetic data curation can convert procedural generators into domain-relevant "data engines," enabling foundation models to improve in data-sparse scientific and industrial domains where real data collection is the primary bottleneck.