Research map / Time series / Causal and generative time series
Modeling Continuous Stochastic Processes with Dynamic Normalizing Flows
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 continuous Wiener base process instead of i.i.d. Gaussian base (ablation) | CTFP · GBM lambda_test=20 | test NLL per observation: 4.094 -> 1.929 | Helped |
| use Wiener base process instead of i.i.d. Gaussian (ablation) | Latent CTFP · Mujoco-Hopper | test NLL per observation: 17.397 -> -31.397 | Helped |
| use Wiener base process instead of i.i.d. Gaussian (ablation) | CTFP · Mujoco-Hopper | test NLL per observation: 22.023 -> -16.249 | Helped |
| add latent variable to CTFP (ablation) | CTFP · Mujoco-Hopper | test NLL per observation: -16.249 -> -31.397 | Helped |
| add latent variable to CTFP (ablation) | CTFP · BAQD | test NLL per observation: -2.361 -> -6.894 | Helped |
| add latent variable to CTFP (ablation) | CTFP · Mixture of OUs | test NLL per observation: 1.408 -> 1.392 | Helped |
| add latent variable to CTFP (ablation) | CTFP · GBM lambda_test=2 | test NLL per observation: 3.107 -> 3.107 | No clear effect |
| replace continuous normalizing flow (ANODE) with RealNVP, no latent (ablation) | CTFP · Mujoco-Hopper | test NLL per observation: -16.249 -> -23.061 | Helped |
| replace continuous normalizing flow with RealNVP, with latent (ablation) | Latent CTFP · Mujoco-Hopper | test NLL per observation: -31.397 -> -23.602 | Hurt |
| use CTFP instead of latent ODE or VRNN on denser test grid | VRNN · GBM lambda_test=20 | test NLL per observation: 3.492 -> 1.929 | Helped |
| use latent CTFP instead of VRNN | VRNN · Mujoco-Hopper | test NLL per observation: 9.113 -> -31.397 | Helped |
| use latent CTFP instead of VRNN | VRNN · PTBDB | test NLL per observation: -1.999 -> -1.999 | No clear effect |
Closest papers by meaning
- Continuous Latent Process Flows (NeurIPS 2021 · similarity 0.87)
- Flexible Tails for Normalizing Flows (ICML 2025 · similarity 0.84)
- Kernelised Normalising Flows (ICLR 2024 · similarity 0.84)
- Probabilistic Forecasting of Irregularly Sampled Time Series with Missing Values via Conditional Normalizing Flows (AAAI 2025 · similarity 0.83)
- Normalizing Flows are Capable Generative Models (ICML 2025 · similarity 0.83)
- Neural Stochastic Flows: Solver-Free Modelling and Inference for SDE Solutions (NeurIPS 2025 · similarity 0.83)
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.