Research map / Time series / Causal and generative time series
Neural Rough Differential Equations for Long Time Series
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 interval log-signatures with NRDE depth 2, step 4 instead of NCDE step 4 | Neural CDE · EigenWorms | accuracy: 66.7 $\pm$ 11.8 -> 83.8 $\pm$ 3.0 | Helped |
| Use NRDE depth 3, step 128 instead of NCDE step 1 | Neural CDE · BIDMC RR | L^2 loss: 2.79 $\pm$ 0.04 -> 1.51 $\pm$ 0.08 | Helped |
| Use NRDE depth 3, step 128 instead of NCDE step 1 | Neural CDE · BIDMC HR | L^2 loss: 9.82 $\pm$ 0.34 -> $\;$2.97 $\pm$ 0.45 | Helped |
| Use NRDE depth 3, step 512 instead of NCDE step 1 | Neural CDE · BIDMC SpO2 | L^2 loss: 2.83 $\pm$ 0.27 -> $\;$1.29 $\pm$ 0.15 | Helped |
| Increase NRDE log-signature depth from 2 to 3 at step 128 (ablation) | Neural RDE · BIDMC RR | L^2 loss: 1.86 $\pm$ 0.03 -> 1.51 $\pm$ 0.08 | Helped |
| Increase NRDE log-signature depth from 2 to 3 at step 128 (ablation) | Neural RDE · BIDMC HR | L^2 loss: 6.77 $\pm$ 0.42 -> $\;$2.97 $\pm$ 0.45 | Helped |
| Increase NRDE log-signature depth from 2 to 3 at step 512 (ablation) | Neural RDE · BIDMC SpO2 | L^2 loss: 2.17 $\pm$ 0.18 -> $\;$1.29 $\pm$ 0.15 | Helped |
| Increase NRDE step size from 8 to 128 at depth 3 (ablation) | Neural RDE · BIDMC RR | L^2 loss: 2.42 $\pm$ 0.19 -> 1.51 $\pm$ 0.08 | Helped |
| Increase NRDE step size from 8 to 128 at depth 3 (ablation) | Neural RDE · BIDMC HR | L^2 loss: 7.67 $\pm$ 0.40 -> $\;$2.97 $\pm$ 0.45 | Helped |
| Increase NRDE step size from 8 to 512 at depth 3 (ablation) | Neural RDE · BIDMC SpO2 | L^2 loss: 2.55 $\pm$ 0.13 -> $\;$1.29 $\pm$ 0.15 | Helped |
| Increase NRDE step size from 128 to 512 at depth 3 (ablation) | Neural RDE · BIDMC HR | L^2 loss: $\;$2.97 $\pm$ 0.45 -> 3.46 $\pm$ 0.13 | Hurt |
| Increase NCDE step size from 1 to 8 (ablation) | Neural CDE · BIDMC RR | L^2 loss: 2.79 $\pm$ 0.04 -> 2.80 $\pm$ 0.06 | No clear effect |
Closest papers by meaning
- Log Neural Controlled Differential Equations: The Lie Brackets Make A Difference (ICML 2024 · similarity 0.91)
- On the Generalization and Approximation Capacities of Neural Controlled Differential Equations (ICLR 2024 · similarity 0.89)
- Stable Neural Stochastic Differential Equations in Analyzing Irregular Time Series Data (ICLR 2024 · similarity 0.89)
- Neural Stochastic Flows: Solver-Free Modelling and Inference for SDE Solutions (NeurIPS 2025 · similarity 0.86)
- Structured Linear CDEs: Maximally Expressive and Parallel-in-Time Sequence Models (NeurIPS 2025 · similarity 0.86)
- Faithful Embeddings of Irregular and Asynchronous Data for Online Log-NCDEs (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.