Research map / Time series / Long-term forecasting
Multivariate Time-Series Forecasting with Temporal Polynomial Graph Neural Networks
Yijing Liu, Qinxian Liu, Jian-Wei Zhang, Haozhe Feng, Zhongwei Wang, Zihan Zhou, Wei 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 |
|---|---|---|---|
| replace TPG-built graph with the dataset prior structure (ablation) | TPGNN · PEMS-D7 horizon 12 | MAE: 3.701 -> 3.215 | Improved |
| replace TPG graph with prior structure (ablation) | TPGNN · PEMS-D7 horizon 12 | RMSE: 7.473 -> 6.559 | Improved |
| add time-varying coefficients instead of static learnable ones (ablation) | TPGNN · PEMS-D7 horizon 12 | MAE: 3.480 -> 3.215 | Improved |
| add average coefficients (overview, Eq. 6) (ablation) | TPGNN · PEMS-D7 horizon 12 | MAE: 3.330 -> 3.215 | Improved |
| add normalization factor (Eq. 7) (ablation) | TPGNN · PEMS-D7 horizon 12 | MAE: 3.261±0.035 -> 3.215±0.014 | No clear effect |
| add multiple parameter matrices (K-matrices) (ablation) | TPGNN · PEMS-D7 horizon 12 | MAPE(%): 8.406±0.245 -> 8.221±0.035 | Improved |
| use TPGNN instead of static/attention graph baselines | TPGNN vs baselines · Synthetic next-step | MFE: 0.1653 -> 0.1266 | Improved |
| use TPGNN instead of Graph Wavenet | TPGNN · Synthetic next-step | MAE: 0.01049 -> 0.01015 | Improved |
| increase polynomial order K from 1 to 4 (ablation) | TPGNN · Synthetic MFE | MFE | Hurt |
| use TPGNN instead of MTGNN on Solar-Energy | TPGNN · Solar-Energy horizon 24 | RSE: 0.4270 -> 0.3498 | Improved |
| use TPGNN instead of MTGNN on Electricity | TPGNN · Electricity horizon 3 | RSE: 0.0745 -> 0.0627 | Improved |
| use TPGNN instead of TPA-LSTM on Exchange-Rate | TPGNN · Exchange-Rate horizon 12 | RSE: 0.0341 -> 0.0350 | No clear effect |
Closest papers by meaning
- TimePFN: Effective Multivariate Time Series Forecasting with Synthetic Data (AAAI 2025 · similarity 0.89)
- Temporal Query Network for Efficient Multivariate Time Series Forecasting (ICML 2025 · similarity 0.89)
- Irregular Multivariate Time Series Forecasting: A Transformable Patching Graph Neural Networks Approach (ICML 2024 · similarity 0.88)
- Fully-Connected Spatial-Temporal Graph for Multivariate Time-Series Data (AAAI 2024 · similarity 0.88)
- Temporal-Aware Evaluation and Learning for Temporal Graph Neural Networks (AAAI 2025 · similarity 0.87)
- Structured Matrix Basis for Multivariate Time Series Forecasting with Interpretable Dynamics (NeurIPS 2024 · 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.
