Research map / Time series / Long-term forecasting
The Forecast After the Forecast: A Post-Processing Shift in Time Series
Daojun Liang, Qi Li, Yinglong Wang, Jing Chen, Hu Zhang, Xiaoxiao Cui, Qizheng Wang, Shuo Li
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 input nudging adapter (Ada-X) to frozen forecaster | Sundial-S · ETTm2 | MSE: 0.348 -> 0.201 | Improved |
| add input nudging adapter (Ada-X) to frozen forecaster | TTM-R2 · ELC | MSE: 0.180 -> 0.167 | Improved |
| add combined input+output adapter (Ada-X+Y) vs SOLID/TAFAS/LoRA baselines | iTransformer · Exchange | MSE: 0.383 -> 0.316 | Improved |
| add TAFAS online adaptation | iTransformer · ETTh1 | MSE: 0.454 -> 0.477 | Hurt |
| add input adapter Ada-X | FourierGNN · Exchange | MSE: 0.380 -> 0.393 | Hurt |
| add Ada-X vs Ada-Y | Autoformer · ETTm1 | MSE: Ada-X 0.604, Ada-Y 0.637 (orig 0.664) | Mixed |
| increase trust-region delta to 0.2 (ablation) | delta-Adapter · iTransformer (figure) | MSE | Hurt |
| use multiplicative instead of additive composite adapter (ablation) | delta-Adapter · PatchTST/TimeMixer | MSE: PatchTST -5.6% vs -5.1%; TimeMixer -1.6% vs -1.8% | No clear effect |
| learn sparse feature mask on inputs | frozen forecaster · ETTh1 | MSE: 0.390 -> 0.382 | Improved |
| fine-tune with LoRA instead of adapter | pre-trained forecaster · time series | MSE | Hurt |
Closest papers by meaning
- TimePerceiver: An Encoder-Decoder Framework for Generalized Time-Series Forecasting (NeurIPS 2025 · similarity 0.88)
- Rethinking Post-Training Recipes for Multimodal Time-Series Forecasting (NeurIPS 2026 · similarity 0.88)
- AdaPTS: Adapting Univariate Foundation Models to Probabilistic Multivariate Time Series Forecasting (ICML 2025 · similarity 0.88)
- Improving Time Series Forecasting via Instance-aware Post-hoc Revision (NeurIPS 2025 · similarity 0.88)
- Enhancing Foundation Models for Time Series Forecasting via Wavelet-based Tokenization (ICML 2025 · similarity 0.88)
- Towards a General Time Series Forecasting Model with Unified Representation and Adaptive Transfer (ICML 2025 · similarity 0.87)
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.
