Research map / Time series / Causal and generative time series
Online time series prediction using feature adjustment
Xiannan Huang, Shuhan Qiu, Jiayuan Du, Chao Yang
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 |
|---|---|---|---|
| Adapt features with ADAPT-Z adapter instead of updating parameters | Base forecasters · ETTm1 | MSE: 0.2211 -> 0.1937 | Improved |
| Feature-space online gradient descent (fOGD) instead of plain OGD | Base forecasters · 13 datasets | MSE: ETTm1 0.2180 -> 0.2178 | Improved |
| Add historical gradient input to adapter (ablation) | ADAPT-Z (SOFTS) · ETTm1 | MSE: 0.1862 -> 0.1857 | Improved |
| Add current feature input to adapter (ablation) | ADAPT-Z (SOFTS) · solar | MSE: 0.0839 -> 0.0791 | Improved |
| Fine-tune base model and adapter on training set, keep online updating (Version1) | ADAPT-Z (iTransformer) · solar | MSE: 0.0940 -> 0.0885 | Improved |
| Freeze all parameters at deployment after training-set finetuning (Version2) | ADAPT-Z (iTransformer) · ETTm1 | MSE: 0.2309 -> 0.2104 | Improved |
| Adapt raw input instead of internal features (ablation) | iTransformer · feature location average | mean MSE: 0.1150 -> 0.1315 | Hurt |
| Use output of first transformer block as features (ablation) | iTransformer · feature location average | mean MSE: 0.1150 -> 0.1041 | Improved |
| Combine DishTS normalization with ADAPT-Z | iTransformer · ETTm1 | MSE: 0.2205 -> 0.1914 | Improved |
| Add DishTS normalization alone | iTransformer · exchange | MSE: 0.0394 -> 0.0792 | Hurt |
| Zero epochs of adapter tuning on validation set (ablation) | ADAPT-Z (SOFTS) · 13 datasets | MSE | Improved |
Closest papers by meaning
- AdaPTS: Adapting Univariate Foundation Models to Probabilistic Multivariate Time Series Forecasting (ICML 2025 · similarity 0.89)
- Learning the Context of Errors: Black-Box Online Adaptation of Time Series Foundation Models (NeurIPS 2026 · similarity 0.87)
- Online Time Series Forecasting with Theoretical Guarantees (NeurIPS 2025 · similarity 0.86)
- 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)
- Lightweight Online Adaption for Time Series Foundation Model Forecasts (ICML 2025 · similarity 0.85)
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.
