Research map / Time series / Causal and generative time series
DAG: A Dual Correlation Network for Time Series Forecasting with Exogenous Variables
Xiangfei Qiu, Yuhan Zhu, Zhengyu Li, Xingjian Wu, Bin Yang, Jilin Hu
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 future exogenous branch G_theta4 to historical endogenous branch G_theta2 (ablation) | DAG · Ablation avg (Electricity, PJM, DE, Energy) | MSE: 0.453 -> 0.273 | Improved |
| add historical endogenous branch G_theta2 to future-exogenous-only model (ablation) | DAG · Ablation avg (Electricity, PJM, DE, Energy) | MSE: 0.444 -> 0.273 | Improved |
| add temporal and channel correlation modules to the two-branch model (ablation) | DAG · Ablation avg (Electricity, PJM, DE, Energy) | MSE: 0.273 -> 0.252 | Improved |
| add temporal and channel correlation modules to the two-branch model (ablation) | DAG · Ablation avg (Electricity, PJM, DE, Energy) | MAE: 0.345 -> 0.331 | Improved |
| add channel correlation on top of G4 branch with G2 (ablation) | DAG · Ablation avg (Electricity, PJM, DE, Energy) | MSE: 0.346 -> 0.252 | Improved |
| add temporal correlation on G2 branch with G4 (ablation) | DAG · Ablation avg (Electricity, PJM, DE, Energy) | MSE: 0.447 -> 0.252 | Improved |
| DAG vs GCGNet on Rapel | DAG · TSF-X Rapel (with future exogenous) | MSE: 0.306 -> 0.230 | Improved |
| DAG vs GCGNet on Sdwpfh2 | DAG · TSF-X Sdwpfh2 (with future exogenous) | MSE: 0.523 -> 0.520 | Hurt |
| DAG vs GCGNet on Sdwpfm2 | DAG · TSF-X Sdwpfm2 (with future exogenous) | MSE: 0.477 -> 0.475 | Hurt |
| DAG vs GCGNet on DE | DAG · TSF-X DE (with future exogenous) | MSE: 0.401 -> 0.370 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- GCGNet — GCGNet: Graph-Consistent Generative Network for Time Series Forecasting with Exogenous Variables (ICLR 2026) · better on 10 of 12 shared results
- TimeXer — TimeXer: Empowering Transformers for Time Series Forecasting with Exogenous Variables (NeurIPS 2024) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TimeXer: Empowering Transformers for Time Series Forecasting with Exogenous Variables (NeurIPS 2024 · similarity 0.89)
- Rethinking Channel Dependence for Multivariate Time Series Forecasting: Learning from Leading Indicators (ICLR 2024 · similarity 0.86)
- Online Time Series Forecasting with Theoretical Guarantees (NeurIPS 2025 · similarity 0.85)
- TimeBridge: Non-Stationarity Matters for Long-term Time Series Forecasting (ICML 2025 · similarity 0.84)
- TiRex-2: Generalizing TiRex to Multivariate Data and Streaming (NeurIPS 2026 · similarity 0.84)
- TimeEmb: A Lightweight Static-Dynamic Disentanglement Framework for Time Series Forecasting (NeurIPS 2025 · similarity 0.84)
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.
