Performance Comparison of CausalMixFT with SMOTE and GANs in Tabular Foundation Model Fine-Tuning on TabMWP Low-Resource Splits

Fine-tuning tabular foundation models (TFMs) under data scarcity is challenging, as early stopping on even scarcer validation data often fails to capture true generalization performance. We propose CausalMixFT, a method that enhances fine-tuning robustness and downstream performance by generating structurally consistent synthetic samples using Structural Causal Models (SCMs) fitted on the target dataset. This approach augments limited real data with causally informed synthetic examples, preserving feature dependencies while expanding training diversity. Evaluated across 33 classification datas

Research goal: How does the performance of CausalMixFT compare to other data augmentation techniques (e.g., SMOTE, GANs) when fine-tuning tabular foundation models on the TabMWP dataset, measured by accuracy on low-resource splits?

Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 8.3/10.