English

Feature-aware Modulation for Learning from Temporal Tabular Data

Machine Learning 2025-12-04 v1

Abstract

While tabular machine learning has achieved remarkable success, temporal distribution shifts pose significant challenges in real-world deployment, as the relationships between features and labels continuously evolve. Static models assume fixed mappings to ensure generalization, whereas adaptive models may overfit to transient patterns, creating a dilemma between robustness and adaptability. In this paper, we analyze key factors essential for constructing an effective dynamic mapping for temporal tabular data. We discover that evolving feature semantics-particularly objective and subjective meanings-introduce concept drift over time. Crucially, we identify that feature transformation strategies are able to mitigate discrepancies in feature representations across temporal stages. Motivated by these insights, we propose a feature-aware temporal modulation mechanism that conditions feature representations on temporal context, modulating statistical properties such as scale and skewness. By aligning feature semantics across time, our approach achieves a lightweight yet powerful adaptation, effectively balancing generalizability and adaptability. Benchmark evaluations validate the effectiveness of our method in handling temporal shifts in tabular data.

Keywords

Cite

@article{arxiv.2512.03678,
  title  = {Feature-aware Modulation for Learning from Temporal Tabular Data},
  author = {Hao-Run Cai and Han-Jia Ye},
  journal= {arXiv preprint arXiv:2512.03678},
  year   = {2025}
}

Comments

17 pages, 6 figures, 8 tables. NeurIPS 2025

R2 v1 2026-07-01T08:07:31.972Z