Research map / Time series / Causal and generative time series
Time Series Forecasting Through the Lens of Dynamics
Alexis-Raja Brachet, Pierre-Yves Richard, CELINE HUDELOT
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 a linear DYN layer (full learnable dynamics) to Informer | Informer · TFB (25 datasets) | average score normalized by NLinear: -0.333 -> -0.228 | Improved |
| add a linear DYN layer to FiLM | FiLM · TFB (25 datasets) | average score normalized by NLinear: -0.036 -> 0.006 | Improved |
| add a linear DYN layer to MICN via data flow update | MICN · TFB (25 datasets) | average score normalized by NLinear: -0.176 -> -0.164 | Improved |
| add a linear DYN layer to FEDformer | FEDformer · TFB (25 datasets) | average score normalized by NLinear: -0.398 -> -0.360 | Improved |
| move a linear DYN layer to the model entry (post-processing configuration) of PatchTST and Crossformer (ablation) | PatchTST, Crossformer · TFB (25 datasets) | MSE/MAE vs vanilla | Hurt |
| move a linear DYN layer to the model entry of iTransformer (ablation) | iTransformer · TFB (25 datasets) | MSE/MAE vs vanilla: better or equal 51% of the time | Mixed |
| replace the DYN layer by a time-preserving linear PRO layer in Informer (control for added parameters) (ablation) | Informer · TFB (25 datasets) | MSE/MAE | Improved |
| use DYN instead of PRO layer in FEDformer (ablation) | FEDformer · TFB (25 datasets) | MSE/MAE | Improved |
| use DYN instead of PRO layer in FiLM (ablation) | FiLM · TFB (25 datasets) | MSE/MAE | No clear effect |
Closest papers by meaning
- LETS Forecast: Learning Embedology for Time Series Forecasting (ICML 2025 · similarity 0.89)
- Time-MoE: Billion-Scale Time Series Foundation Models with Mixture of Experts (ICLR 2025 · similarity 0.87)
- Scaling Law for Time Series Forecasting (NeurIPS 2024 · similarity 0.87)
- SynTSBench: Rethinking Temporal Pattern Learning in Deep Learning Models for Time Series (NeurIPS 2025 · similarity 0.87)
- TEMPO: Prompt-based Generative Pre-trained Transformer for Time Series Forecasting (ICLR 2024 · similarity 0.86)
- A Closer Look at Transformers for Time Series Forecasting: Understanding Why They Work and Where They Struggle (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.
