Research map / Probabilistic and scientific modeling / Dynamical systems
DAISI: Data Assimilation with Inverse Sampling using Stochastic Interpolants
Martin Andrae, Erik Wikingsson, So Takao, Tomas Landelius, Fredrik Lindsten
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 inverse sampling of forecast ensemble before guided sampling (ablation) | DAISI · L63 | RMSE: 5.90 -> 2.03 | Improved |
| tune tmin (instead of tmin near 0) with eps=0 (ablation) | DAISI · L63 | RMSE: 7.04 -> 2.44 | Improved |
| tmin near 0 and eps=0 without tuning (ablation) | DAISI · L63 | CRPS: 7.17 (no inversion) -> 11.9 | Hurt |
| add nonzero eps in addition to tuned tmin (ablation) | DAISI · L63 | SSR: 0.32 -> 1.08 | Improved |
| add nonzero eps in addition to tuned tmin (ablation) | DAISI · L63 | CRPS: 3.19 -> 2.61 | Improved |
| add nonzero eps in addition to tuned tmin (ablation) | DAISI · L63 | RMSE: 2.05 -> 2.03 | No clear effect |
| intermediate tmin (0.3) versus 0.01 or 0.6 (ablation) | DAISI · 1D toy problem | MMD: 0.15 / 0.29 -> 0.03 | Improved |
| use DAISI instead of FlowDAS filtering | SQG forecast model · SQG Sparse | CRPS: 3.34 -> 1.73 | Improved |
| replace numerical model with FlowDAS learned forecast model (DAISI-ML) (ablation) | DAISI · SQG Sparse | CRPS: 1.73 -> 1.72 | No clear effect |
| DAISI vs LETKF under saturating arctan observations | SQG Saturating · SQG Saturating | CRPS: 5.24 -> 1.54 | Improved |
| DAISI vs EnSF under saturating observations | SQG Saturating · SQG Saturating | CRPS: 1.33 (EnSF) vs 1.54 | No clear effect |
| DAISI vs LETKF with noisy linear observations | SQG Noisy · SQG Noisy | CRPS: 1.34 -> 1.32 | 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 5 of 5 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
- FlowDAS: A Stochastic Interpolant-based Framework for Data Assimilation (NeurIPS 2025 · similarity 0.91)
- Deep Bayesian Filter for Bayes-Faithful Data Assimilation (ICML 2025 · similarity 0.89)
- Learning semilinear neural operators: A unified recursive framework for prediction and data assimilation. (ICLR 2024 · similarity 0.86)
- Implicit Maximum a Posteriori Filtering via Adaptive Optimization (ICLR 2024 · similarity 0.86)
- On conditional diffusion models for PDE simulations (NeurIPS 2024 · similarity 0.86)
- Nonlinear Filtering with Brenier Optimal Transport Maps (ICML 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.
