Research map / Architectures and efficiency / RNNs and normalization
Efficient and Accurate Gradients for Neural SDEs
Patrick Kidger, James Foster, Xuechen Li, Terry Lyons
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 midpoint with reversible Heun (ablation) | SDE-GAN · Weights | MMD (x1e-2): 4.38 -> 1.75 | Improved |
| replace midpoint with reversible Heun (ablation) | SDE-GAN · Weights | training time (days): 5.12 -> 2.59 | Improved |
| replace midpoint with reversible Heun (ablation) | Latent SDE · Air quality | label classification accuracy (%): 46.3 -> 49.2 | Improved |
| replace midpoint with reversible Heun (ablation) | Latent SDE · Air quality | MMD (x1e-2): 0.591 -> 0.472 | Improved |
| replace midpoint with reversible Heun (ablation) | Latent SDE · Air quality | training time (hours): 5.58 -> 4.47 | Improved |
| replace Virtual Brownian Tree with Brownian Interval (ablation) | SDE solver · SDE solve, size 32768 | time (s): 5.00 x 10^2 -> 4.7 x 10^1 | Improved |
| replace Virtual Brownian Tree with Brownian Interval (ablation) | SDE solver · SDE solve, size 1 | time (s): 1.6 -> 8.2 x 10^-1 | Improved |
| replace Virtual Brownian Tree with Brownian Interval (ablation) | SDE solver · doubly sequential access, size 32768 | time (s): 2.9 x 10^0 -> 3.5 x 10^-1 | Improved |
| replace gradient penalty with weight clipping (ablation) | SDE-GAN with midpoint · OU dataset | MMD (x1e-1): 2.58 -> 1.03 | Improved |
| replace gradient penalty with weight clipping (ablation) | SDE-GAN with midpoint · OU dataset | training time (hours): 55.0 -> 32.5 | Improved |
| replace gradient penalty with weight clipping (ablation) | SDE-GAN with midpoint · OU dataset | real/fake classification accuracy (%): 98.2 -> 93.9 | Improved |
| replace midpoint with reversible Heun under clipping (ablation) | SDE-GAN · OU dataset | MMD (x1e-1): 1.03 -> 0.45 | Improved |
Closest papers by meaning
- SING: SDE Inference via Natural Gradients (NeurIPS 2025 · similarity 0.88)
- Efficient Training of Neural Stochastic Differential Equations by Matching Finite Dimensional Distributions (ICLR 2025 · similarity 0.88)
- Neural Stochastic Flows: Solver-Free Modelling and Inference for SDE Solutions (NeurIPS 2025 · similarity 0.87)
- Exact, Tractable Gauss-Newton Optimization in Deep Reversible Architectures Reveal Poor Generalization (NeurIPS 2024 · similarity 0.86)
- Adaptive Methods through the Lens of SDEs: Theoretical Insights on the Role of Noise (ICLR 2025 · similarity 0.86)
- Stable Neural Stochastic Differential Equations in Analyzing Irregular Time Series Data (ICLR 2024 · similarity 0.85)
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.
