Research map / Time series / Causal and generative time series
Deep Time Series Forecasting With Shape and Temporal Criteria
Vincent Le Guen, Nicolas Thome
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 |
|---|---|---|---|
| train with DILATE instead of MSE loss | Seq2Seq GRU · ECG5000 | TDI (x10): 8.27 -> 6.59 | Improved |
| train with DILATE instead of MSE loss | Seq2Seq GRU · Traffic | TDI (x100): 15.4 -> 14.4 | Improved |
| train with DILATE instead of MSE loss | MLP · Synthetic-det | DTW (x10): 38.6 -> 32.1 | Improved |
| train with DILATE instead of MSE loss | Seq2Seq GRU · ECG5000 | MSE (x100): 21.2 -> 30.3 | Hurt |
| train with DILATE instead of soft-DTW loss | Seq2Seq GRU · ECG5000 | TDI (x10): 27.2 -> 6.59 | Improved |
| train with DILATE instead of soft-DTW loss | Seq2Seq GRU · Synthetic-det | DTW (x10): 22.7 -> 23.1 | No clear effect |
| train with DILATE instead of MSE loss | N-Beats · ETTh1 (96) | DILATE: 7.4 -> 4.6 | Improved |
| train with DILATE instead of MSE loss | Informer · Synthetic | DILATE: 16.6 -> 15.1 | Improved |
| add quality constraint via posterior network (STRIPE++ over STRIPE) | STRIPE · Synthetic-prob (multiple futures) | CRPS (x1000): 60.5 -> 48.6 | Improved |
| add quality constraint via posterior network (STRIPE++ over STRIPE) | STRIPE · Synthetic-prob (multiple futures) | H_quality (DILATE): 15.5 -> 15.0 | Improved |
| use DILATE as the diversity kernel instead of no diversity loss (ablation) | cVAE DILATE · Synthetic-prob (multiple futures) | H_diversity (DILATE): 33.9 -> 31.3 | No clear effect |
| add shape kernel diversity loss (ablation) | cVAE DILATE · Synthetic-prob (multiple futures) | H_diversity (DTW): 18.8 -> 16.4 | Improved |
Closest papers by meaning
- SynTSBench: Rethinking Temporal Pattern Learning in Deep Learning Models for Time Series (NeurIPS 2025 · similarity 0.88)
- Loss Shaping Constraints for Long-Term Time Series Forecasting (ICML 2024 · similarity 0.88)
- ProbTS: Benchmarking Point and Distributional Forecasting across Diverse Prediction Horizons (NeurIPS 2024 · similarity 0.87)
- TimeBase: The Power of Minimalism in Efficient Long-term Time Series Forecasting (ICML 2025 · similarity 0.87)
- TimeEmb: A Lightweight Static-Dynamic Disentanglement Framework for Time Series Forecasting (NeurIPS 2025 · similarity 0.86)
- FilterNet: Harnessing Frequency Filters for Time Series Forecasting (NeurIPS 2024 · 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.
