Research map / Time series / Causal and generative time series
Conditionally Whitened Generative Models for Probabilistic Time Series Forecasting
Yanfeng Yang, Siwei Chen, Pingping Hu, Zhaotong Shen, Yingjie Zhang, Zhuoran Sun, Shuai Li, Ziqi Chen, Kenji Fukumizu
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 conditional whitening (CW-Gen) with JMCE prior | TimeDiff · ETTh1 | Conditional FID: 19.008 -> 6.788 | Improved |
| add conditional whitening (CW-Gen) with JMCE prior | SSSD · ETTh1 | CRPS: 0.836 -> 0.524 | Improved |
| add conditional whitening (CW-Gen) with JMCE prior | Diffusion-TS · ETTh1 | Conditional FID: 81.563 -> 7.686 | Improved |
| add conditional whitening (CW-Gen) with JMCE prior | FlowTS · ETTh1 | CRPS: 0.724 -> 0.488 | Improved |
| add conditional whitening to a prior-informed model | TMDM · ETTh1 | QICE: 3.360 -> 4.555 | Hurt |
| add conditional whitening to a prior-informed model | NsDiff · ETTh1 | CRPS: 0.407 -> 0.431 | Hurt |
| add conditional whitening to a prior-informed model | NsDiff · ETTh1 | QICE: 1.792 -> 1.249 | Improved |
| add conditional whitening across six generative models | six models · ETTh1 | win rate over all metrics: CRPS/QICE 16.7% -> 83.3%; ProbCorr/Cond. FID 0.0% -> 100.0% | Improved |
| add conditional whitening on five datasets | CW-Gen · five datasets | win rate vs original method: ETTh1 22/24; ILI 20/24; Solar Energy 19/24 | Improved |
| add conditional whitening | CW-Gen · five datasets | ProbMSE: 76.67% of model-dataset combinations | Improved |
| add conditional whitening | CW-Gen · five datasets | ProbMAE: 80.00% of model-dataset combinations | Improved |
Closest papers by meaning
- Flow Matching with Gaussian Process Priors for Probabilistic Time Series Forecasting (ICLR 2025 · similarity 0.90)
- Conditional Diffusion Model with Nonlinear Data Transformation for Time Series Forecasting (ICML 2025 · similarity 0.89)
- Probabilistic Time Series Modeling with Decomposable Denoising Diffusion Model (ICML 2024 · similarity 0.87)
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (NeurIPS 2025 · similarity 0.87)
- Probabilistic Forecasting of Irregularly Sampled Time Series with Missing Values via Conditional Normalizing Flows (AAAI 2025 · similarity 0.87)
- HDT: Hierarchical Discrete Transformer for Multivariate Time Series Forecasting (AAAI 2025 · 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.
