Research map / Time series / Long-term forecasting
Reviving Error Correction in Modern Deep Time-Series Forecasting
Minh Hoang Nguyen, Van Dai Do, Huu Hiep Nguyen, Dung Nguyen, Kien Do, Hung Le
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 UEC-STD error corrector to autoregressive backbones | AR forecasters (TimeMixer, TimesNet, TimeXer) · Seven common datasets | MSE: 2.32% | Improved |
| add UEC-STD to autoregressive backbones | AR forecasters · Monash datasets | MSE: 10.05% | Improved |
| use UEC-STD instead of UEC-MLP corrector | AR forecasters · ETTm2 | MSE error reduction (%): -2.71 -> -4.23 | Improved |
| use LSTM corrector instead of no correction | AR forecasters · ETTm2 | MSE error reduction (%): 0.00 -> 16.64 | Hurt |
| use tree/logistic classical correctors on large datasets | AR forecasters · Traffic and Electricity | training convergence | Hurt |
| add trend and seasonal inputs only (no STD output) (ablation) | UEC corrector · ETTh1 | MSE: 0.451 -> 0.446 | Improved |
| add seasonal output to trend-only output (ablation) | UEC-STD · ETTh1 | MSE: 0.464 -> 0.424 | Improved |
| add trend output to seasonal-only output (ablation) | UEC-STD · ETTh1 | MSE: 0.427 -> 0.424 | Improved |
| add full STD input and output to plain corrector (ablation) | UEC · Traffic | MSE: 0.545 -> 0.543 | No clear effect |
| add full STD input and output to plain corrector (ablation) | UEC · Weather | MSE: 0.241 -> 0.239 | Improved |
| use balanced beta validation (mix training data) instead of validation-only (ablation) | UEC-STD on TimesNet · ETTh1 | MSE: 0.505 -> 0.450 | Improved |
| use balanced beta validation instead of validation-only (ablation) | UEC-STD · ETTh1 and Weather, two backbones | MSE/MAE: 6/8 settings | Improved |
Closest papers by meaning
- Multivariate Probabilistic Time Series Forecasting with Correlated Errors (NeurIPS 2024 · similarity 0.87)
- SynTSBench: Rethinking Temporal Pattern Learning in Deep Learning Models for Time Series (NeurIPS 2025 · similarity 0.87)
- U-Mixer: An Unet-Mixer Architecture with Stationarity Correction for Time Series Forecasting (AAAI 2024 · similarity 0.87)
- Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts (ICLR 2025 · similarity 0.86)
- Unified Training of Universal Time Series Forecasting Transformers (ICML 2024 · similarity 0.86)
- LETS Forecast: Learning Embedology for Time Series Forecasting (ICML 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.
