Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Stochastic Differential Equations with Variational Wishart Diffusions
Martin Jørgensen, Marc Deisenroth, Hugh Salimbeni
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 |
|---|---|---|---|
| Replace diagonal diffusion with Wishart diffusion | diffGP flow · UCI regression (Concrete, kin8nm, Naval, Protein) | test-set log-likelihood: -3.0401± 0.1460 -> -3.0164± 0.1352; 1.3046± 0.0302 -> 1.3123± 0.0305; 7.2256± 0.0949 -> 8.2736± 0.0677; -2.8599± 0.0109 | Improved |
| Replace diagonal diffusion with Wishart diffusion | diffGP flow · UCI regression (Boston) | test-set log-likelihood: -2.7586± 0.7301 -> -3.1014± 0.8984 | Hurt |
| Replace diagonal diffusion with Wishart diffusion | diffGP flow · UCI regression (Bike, Power, Wine_white) | test-set log-likelihood: -0.6842± 0.1143 -> -0.6569± 0.0381; -2.7753± 0.0402 -> -2.7736± 0.0413; -1.0616± 0.0448 -> -1.0603± 0.0443 | No clear effect |
| Replace diagonal diffusion with Wishart diffusion | diffGP flow · UCI regression (Concrete, kin8nm, Naval, Protein) | test-set RMSE: 4.9913± 0.6657 -> 4.9086± 0.6442; 0.0646± 0.0021 -> 0.0642± 0.0023; <0.0001± <0.0001 -> <0.0001± <0.0001; 4.1717± 0.0479 | Improved |
| Replace diagonal diffusion with Wishart diffusion | diffGP flow · UCI regression (Boston) | test-set RMSE: 2.9628± 0.6991 -> 3.1713± 0.8296 | Hurt |
| Use deterministic flow with no diffusion (ablation) | GP flow · UCI regression | test-set log-likelihood | Hurt |
| Remove drift (ablation) | diffWGP · Beijing air-quality forecasting | forecast log-likelihood | Mixed |
| Restrict diffusion to diagonal covariance (ablation) | diffWGP · Beijing air-quality forecasting | forecast log-likelihood | Hurt |
| Vary Wishart rank from rho = 5 to rho = 10 (ablation) | diffWGP · UCI regression | test-set log-likelihood | No clear effect |
Closest papers by meaning
- Statistical Spatially Inhomogeneous Diffusion Inference (AAAI 2024 · similarity 0.86)
- Variational Inference for SDEs Driven by Fractional Noise (ICLR 2024 · similarity 0.86)
- Dynamical Diffusion: Learning Temporal Dynamics with Diffusion Models (ICLR 2025 · similarity 0.86)
- Stability and Generalizability in SDE Diffusion Models with Measure-Preserving Dynamics (NeurIPS 2024 · similarity 0.86)
- Neural Stochastic Flows: Solver-Free Modelling and Inference for SDE Solutions (NeurIPS 2025 · similarity 0.86)
- ABC: Any-Subset Autoregression via Non-Markovian Diffusion Bridges in Continuous Time and Space (NeurIPS 2026 · 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.
