Research map / Image generation / GANs
Probabilistic Forecasting with Generative Networks via Scoring Rule Minimization
Lorenzo Pacchiardi, Rilwan A. Adewoyin, Peter Dueben, Ritabrata Dutta
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 Energy-Kernel Score instead of WGAN-GP (3) | generative forecaster · Lorenz96 | calibration error: 0.1648 -> 0.0104 | Improved |
| train with Energy Score instead of GAN (3) | generative forecaster · Lorenz63 | NRMSE: 0.0698 -> 0.0105 | Improved |
| train with Kernel Score instead of regression loss | GRU forecaster · Lorenz96 | NRMSE: 0.0198 -> 0.0164 | Improved |
| train with Kernel Score instead of regression loss | GRU forecaster · Lorenz63 | NRMSE: 0.0079 -> 0.0083 | No clear effect |
| use Kernel Score instead of Energy Score | GRU forecaster · Lorenz96 | calibration error: 0.0205 -> 0.2196 | Hurt |
| use Energy-Kernel Score instead of Energy Score | GRU forecaster · Lorenz63 | calibration error: 0.0380 -> 0.1000 | Hurt |
| pick GAN config (2) for calibration instead of (1) for deterministic skill | GAN · Lorenz63 | NRMSE: 0.0274 -> 0.2425 | Hurt |
| pick GAN config (1) for deterministic skill instead of (2) | GAN · Lorenz63 | calibration error: 0.0860 -> 0.4830 | Hurt |
| use Patched Energy Score (patch 8) instead of Energy Score | U-Net forecaster · WeatherBench Z500 | calibration error: 0.0863 -> 0.0550 | Improved |
| add patching to Energy Score (patch 16 vs unpatched) | U-Net forecaster · WeatherBench Z500 | NRMSE: 0.1208 -> 0.1186 | Improved |
| combine Kernel Score with Variogram Score instead of Kernel alone | U-Net forecaster · WeatherBench Z500 | calibration error: 0.0797 -> 0.1704 | Hurt |
| combine Energy with Kernel Score instead of Energy alone | U-Net forecaster · WeatherBench Z500 | NRMSE: 0.1208 -> 0.1194 | Improved |
Closest papers by meaning
- Adaptive Quantization in Generative Flow Networks for Probabilistic Sequential Prediction (NeurIPS 2025 · similarity 0.88)
- Probabilistic Forecasting with Stochastic Interpolants and Föllmer Processes (ICML 2024 · similarity 0.87)
- Adversarial Attacks on Probabilistic Autoregressive Forecasting Models (ICML 2020 · similarity 0.87)
- Time-series Generative Adversarial Networks (NeurIPS 2019 · similarity 0.87)
- Koopman Generative Operators for Efficient Probabilistic Time-Series Forecasting (NeurIPS 2026 · similarity 0.86)
- Time Series Generation Under Data Scarcity: A Unified Generative Modeling Approach (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.
