Research map / Time series / Causal and generative time series
Deep Rao-Blackwellised Particle Filters for Time Series Forecasting
Richard Kurle, Syama Sundar Rangapuram, Emmanuel de Bézenac, Stephan Günnemann, Jan Gasthaus
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 state-to-switch recurrence and auxiliary-variable non-linear emission (ARSGLS vs KVAE-RB, same GLS and emission) | switching GLS with Rao-Blackwellised particle filter · wiki rolling | CRPS: 0.245 -> 0.217 | Improved |
| ARSGLS instead of KVAE-RB | deep state space forecasting · electricity rolling | CRPS: 0.296 -> 0.138 | Improved |
| RSGLS-ISSM instead of DeepState (same ISSM structure) | ISSM state space forecasting · exchange rolling | CRPS: 0.010 -> 0.007 | Improved |
| RSGLS-ISSM instead of DeepState | ISSM state space forecasting · traffic rolling | CRPS: 0.131 -> 0.148 | Hurt |
| RSGLS-ISSM instead of DeepState | ISSM state space forecasting · electricity rolling | CRPS: 0.071 -> 0.070 | No clear effect |
| add state-to-switch dependence to SGLS (RSGLS vs SGLS) (ablation) | switching GLS · synthetic damped pendulum | forecast quality (qualitative) | Improved |
| ARSGLS instead of KVAE-MC/KVAE-RB | video forecasting · Box, Box-Gravity, Polygon, Pong | Wasserstein distance | Improved |
| use Rao-Blackwellised objective instead of Monte Carlo smoothing samples (KVAE-RB vs KVAE-MC) (ablation) | KVAE · wiki rolling | CRPS: 0.276 -> 0.245 | Improved |
Closest papers by meaning
- FilterNet: Harnessing Frequency Filters for Time Series Forecasting (NeurIPS 2024 · similarity 0.86)
- Probabilistic Forecasting with Stochastic Interpolants and Föllmer Processes (ICML 2024 · similarity 0.86)
- Shifting Time: Time-series Forecasting with Khatri-Rao Neural Operators (ICML 2025 · similarity 0.85)
- $K^2$VAE: A Koopman-Kalman Enhanced Variational AutoEncoder for Probabilistic Time Series Forecasting (ICML 2025 · similarity 0.85)
- Nonlinear Filtering with Brenier Optimal Transport Maps (ICML 2024 · similarity 0.84)
- Implicit Maximum a Posteriori Filtering via Adaptive Optimization (ICLR 2024 · 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.
