Research map / Probabilistic and scientific modeling / Dynamical systems
Training-Free Bayesian Filtering with Generative Emulators
Thomas Savary, François Rozet, Gilles Louppe
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 |
|---|---|---|---|
| use optimal proposal from diffusion emulator (FA-APF) instead of bootstrap proposal | particle filter · Lorenz63 observe x and z, N=256 | skill: 1.25 -> 0.6 | Improved |
| use optimal proposal with weighting instead of bootstrap proposal | particle filter · Lorenz63 observe x only, N=256 | skill: 2.38 -> 1.87 | Improved |
| FA-APF instead of EnKF | ensemble filter · Lorenz63 observe x and z, N=256 | skill: 0.72 -> 0.6 | Improved |
| FA-APF instead of EnKF with only first component observed | ensemble filter · Lorenz63 observe x only, N=256 | skill: 2.79 -> 1.87 | Improved |
| FA-APF instead of EnSF | ensemble filter · Lorenz63 observe x only, N=256 | skill: 3.76 -> 1.87 | Improved |
| FA-APF instead of bootstrap particle filter | particle filter · Navier-Stokes coarse (8,8) | skill: 3.02 -> 2.39 | Improved |
| FA-APF instead of FlowDAS (single-particle fully adapted filter extended to ensemble) | ensemble filter · Navier-Stokes sparse (16,16) | skill: 1.9 -> 1.12 | Improved |
| FA-APF instead of FlowDAS | ensemble filter · Navier-Stokes coarse (32,32) | skill: 0.19 -> 0.14 | Improved |
| FA-APF instead of FlowDAS | ensemble filter · Navier-Stokes sparse (8,8) | skill: 3.08 -> 2.30 | Improved |
| increase observation resolution from (8,8) to (32,32) coarse | FA-APF · Navier-Stokes coarse | skill: 2.39 -> 0.14 | Improved |
| use approximate Dirac-mean weights instead of exact Monte Carlo weights (ablation) | FA-APF · Lorenz63 | skill | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FlowDAS — FlowDAS: A Stochastic Interpolant-based Framework for Data Assimilation (NeurIPS 2025) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Implicit Maximum a Posteriori Filtering via Adaptive Optimization (ICLR 2024 · similarity 0.88)
- Nonlinear Filtering with Brenier Optimal Transport Maps (ICML 2024 · similarity 0.86)
- Thermalizer: Stable autoregressive neural emulation of spatiotemporal chaos (ICML 2025 · similarity 0.85)
- Predicting partially observable dynamical systems via diffusion models with a multiscale inference scheme (NeurIPS 2025 · similarity 0.84)
- Lost in Latent Space: An Empirical Study of Latent Diffusion Models for Physics Emulation (NeurIPS 2025 · similarity 0.84)
- Nesting Particle Filters for Experimental Design in Dynamical Systems (ICML 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.
