Signals, Concepts, and Laws: Toward Universal, Explainable Time-Series Forecasting

Hongwei Ma; Junbin Gao; Minh-Ngoc Tran

Signals, Concepts, and Laws: Toward Universal, Explainable Time-Series Forecasting

Machine Learning 2025-12-02 v3

Authors: Hongwei Ma , Junbin Gao , Minh-Ngoc Tran

Abstract

Accurate, explainable and physically credible forecasting remains a persistent challenge for multivariate time-series whose statistical properties vary across domains. We propose DORIC, a Domain-Universal, ODE-Regularized, Interpretable-Concept Transformer for Time-Series Forecasting that generates predictions through five self-supervised, domain-agnostic concepts while enforcing differentiable residuals grounded in first-principles constraints.

Keywords

time series forecasting

Cite

@article{arxiv.2508.01407,
  title  = {Signals, Concepts, and Laws: Toward Universal, Explainable Time-Series Forecasting},
  author = {Hongwei Ma and Junbin Gao and Minh-Ngoc Tran},
  journal= {arXiv preprint arXiv:2508.01407},
  year   = {2025}
}

Related papers

View all related →