Research map / Time series / Long-term forecasting
SparseTSF: Lightweight and Robust Time Series Forecasting via Sparse Modeling
Shengsheng Lin, Weiwei Lin, Wentai Wu, Haojun Chen, C L Philip Chen
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 Sparse technique (cross-period downsampling) (ablation) | Linear · ETTh1 (H=720) | MSE: 0.424 -> 0.417 | Improved |
| add Sparse technique (ablation) | Transformer · ETTh1 (H=720) | MSE: 0.770 -> 0.489 | Improved |
| add Sparse technique (ablation) | GRU · ETTh1 (H=720) | MSE: 0.620 -> 0.455 | Improved |
| add Sparse technique (ablation) | Linear · ETTh2 (H=96) | MSE: 0.257 -> 0.267 | Hurt |
| set period w to the true period 24 instead of 6 (ablation) | SparseTSF · ETTh1 (avg) | MSE: 0.405 -> 0.394 | Improved |
| set w to 48 instead of 24 (ablation) | SparseTSF · ETTh1 (avg) | MSE: 0.394 -> 0.402 | Hurt |
| use large w=144 on Weather instead of w=2 (ablation) | SparseTSF · Weather (H=720) | MSE: 0.318 -> 0.332 | Hurt |
| use sparse w=6 instead of fully connected w=1 (ablation) | SparseTSF · ETTm1 (H=720) | MSE: 0.429 -> 0.415 | Improved |
| add instance normalization (ablation) | SparseTSF · ETTh2 (H=720) | MSE: 0.866 -> 0.370 | Improved |
| add instance normalization (ablation) | SparseTSF · Electricity (H=720) | MSE: 0.198 -> 0.203 | No clear effect |
| lengthen look-back from 96 to 720 (ablation) | SparseTSF · Electricity (avg) | MSE: 0.222 -> 0.163 | Improved |
| train on Electricity, test on ETTh1 (cross-domain) with SparseTSF instead of FITS | SparseTSF · Electricity to ETTh1 (H=96) | MSE: 0.380 -> 0.373 | Improved |
Closest papers by meaning
- SparseTSF: Modeling Long-term Time Series Forecasting with *1k* Parameters (ICML 2024 · similarity 0.99)
- TimeBase: The Power of Minimalism in Efficient Long-term Time Series Forecasting (ICML 2025 · similarity 0.88)
- Parsimony or Capability? Decomposition Delivers Both in Long-term Time Series Forecasting (NeurIPS 2024 · similarity 0.87)
- A Lightweight Sparse Interaction Network for Time Series Forecasting (AAAI 2025 · similarity 0.86)
- Unlocking the Power of Patch: Patch-Based MLP for Long-Term Time Series Forecasting (AAAI 2025 · similarity 0.86)
- Time-LLM: Time Series Forecasting by Reprogramming Large Language Models (ICLR 2024 · 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.
