Research map / Time series / Causal and generative time series
Neural Flows: Efficient Alternative to Neural ODEs
Marin Biloš, Johanna Sommer, Syama Sundar Rangapuram, Tim Januschowski, Stephan Günnemann
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 the ODE solver with a coupling flow | Neural ODE smoothing model · MuJoCo | MSE: 8.403{ $\pm$0.142} -> 4.217{ $\pm$0.147} | Improved |
| Replace the ODE solver with a ResNet flow | Neural ODE smoothing model · Activity | MSE: 6.390{ $\pm$0.136} -> 6.279{ $\pm$0.098} | Improved |
| Replace the ODE solver with a coupling flow | Neural ODE smoothing model · Physionet | MSE: 4.833{ $\pm$0.078} -> 4.860{ $\pm$0.070} | No clear effect |
| Replace the ODE solver with a coupling flow | Neural ODE smoothing model · Physionet | AUC: 0.777{ $\pm$0.012} -> 0.788{ $\pm$0.004} | Improved |
| Replace GRU-ODE with GRU flow | GRU-ODE forecasting model · MIMIC-III | MSE: 0.507{ $\pm$0.005} -> 0.499{ $\pm$0.004} | Improved |
| Replace GRU-ODE with GRU flow | GRU-ODE forecasting model · MIMIC-III | NLL: 0.770{ $\pm$0.023} -> 0.781{ $\pm$0.041} | Hurt |
| Replace GRU-ODE with GRU flow | GRU-ODE forecasting model · MIMIC-IV | MSE: 0.379{ $\pm$0.005} -> 0.364{ $\pm$0.008} | Improved |
| Replace GRU-ODE with GRU flow | GRU-ODE forecasting model · MIMIC-IV | NLL: 0.748{ $\pm$0.045} -> 0.734{ $\pm$0.054} | Improved |
| Use a ResNet flow instead of a discrete GRU | Temporal point process model · MOOC | NLL: -0.4448 -> -1.2379 | Improved |
| Use a ResNet flow instead of a discrete GRU | Temporal point process model · Reddit | NLL: -0.9299 -> -1.2962 | Improved |
| Use a coupling flow instead of Jump ODE | Temporal point process model · Wiki | NLL: -1.3635 -> -1.2702 | Hurt |
| Replace the ODE solver with a ResNet flow | Neural ODE synthetic dynamics model · Ellipse | MSE (× 10^-2): 25.59{ $\pm$3.19} -> 9.48{ $\pm$2.64} | Improved |
Closest papers by meaning
- Neural ODE Processes (ICLR 2021 · similarity 0.90)
- ResNet After All: Neural ODEs and Their Numerical Solution (ICLR 2021 · similarity 0.90)
- Stable Neural Stochastic Differential Equations in Analyzing Irregular Time Series Data (ICLR 2024 · similarity 0.90)
- Neural Controlled Differential Equations for Irregular Time Series (NeurIPS 2020 · similarity 0.89)
- Neural Stochastic Flows: Solver-Free Modelling and Inference for SDE Solutions (NeurIPS 2025 · similarity 0.88)
- How Deep Do We Need: Accelerating Training and Inference of Neural ODEs via Control Perspective (ICML 2024 · similarity 0.88)
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.
