Research map / Time series / Foundation models for time series
UniCA: Unified Covariate Adaptation for Time Series Foundation Model
Lu Han, Yu Liu, Lan Li, Qiwen Deng, Jian Jiang, Yinbo sun, Zhe Yu, Binfeng Wang, Xingyu Lu, Lintao Ma, Han-Jia Ye, De-Chuan Zhan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add UniCA adapter | Chronos-Bolt · 12 unimodal covariate datasets | MAPE: 0.506 MAPE, best among adapter methods | Improved |
| add UniCA instead of specialized methods | TSFMs · 12 unimodal covariate datasets | MAPE: 0.596-0.738 -> 0.506 | Improved |
| standard full fine-tuning of the TSFM | TSFMs · 12 unimodal covariate datasets | MAPE | No clear effect |
| add UniCA adapter | TimesFM · MMSP + Time-MMD | error: 6.5% reduction in error | Improved |
| add UniCA adapter | Chronos-Bolt · MMSP + Time-MMD | error: 5.9% reduction | Improved |
| add Covariate Homogenizer | TFT · MMSP | MAE: 5% drop | Improved |
| add Covariate Homogenizer | TFT · Time-MMD | MAE: 13.0% drop | Improved |
| add Covariate Homogenizer | TiDE · MMSP | MAE: 30.1% reduction | Improved |
| add Covariate Homogenizer | TiDE · Time-MMD | MAE: 10.5% reduction | Improved |
| use UniCA adapter instead of zero-shot Moirai (ablation) | Moirai · 12 unimodal covariate datasets | MAPE, MSE, MAE, CRPS | Improved |
| freeze backbone with adapter instead of full fine-tuning (ablation) | Chronos-Bolt (UniCA) · 12 unimodal covariate datasets | MAPE, CRPS | Improved |
| use Linear homogenizer instead of MLP (ablation) | UniCA · 12 unimodal covariate datasets | relative MAPE | No clear effect |
Closest papers by meaning
- AdaPTS: Adapting Univariate Foundation Models to Probabilistic Multivariate Time Series Forecasting (ICML 2025 · similarity 0.89)
- Less is More: Unlocking Specialization of Time Series Foundation Models via Structured Pruning (NeurIPS 2025 · similarity 0.88)
- Learning the Context of Errors: Black-Box Online Adaptation of Time Series Foundation Models (NeurIPS 2026 · similarity 0.88)
- Learning Pattern-Specific Experts for Time Series Forecasting Under Patch-level Distribution Shift (NeurIPS 2025 · similarity 0.86)
- ARM: Refining Multivariate Forecasting with Adaptive Temporal-Contextual Learning (ICLR 2024 · similarity 0.86)
- TS-RAG: Retrieval-Augmented Generation based Time Series Foundation Models are Stronger Zero-Shot Forecaster (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
