Research map / Time series / Causal and generative time series
Learning the Dynamics of Sparsely Observed Interacting Systems
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 SigLasso instead of the strongest baseline, Neural CDE | Full-path reconstruction in the well-specified setting · Well-specified | L2 error: 0.61 ± 0.38 -> 0.13 ± 0.07 | Helped |
| Use SigLasso instead of the strongest baseline, Neural CDE | Last-point prediction in the well-specified setting · Well-specified | MSE on last point: 1.46 ± 1.20 -> 0.73 ± 0.56 | Helped |
| Use SigLasso instead of the strongest baseline, GRU | Full-path reconstruction in the ill-specified setting · Ill-specified | L2 error: 0.24 ± 0.11 -> 0.15 ± 0.02 | Helped |
| Use SigLasso instead of the strongest baseline, GRU | Last-point prediction in the ill-specified setting · Ill-specified | MSE on last point: 0.19 ± 0.09 -> 0.09 ± 0.05 | Helped |
| Use SigLasso instead of the strongest baseline, GRU | Full-path reconstruction in the OU setting · OU | L2 error: 0.05 ± 0.06 -> 0.01 ± 0.02 | Helped |
| Use SigLasso instead of the strongest baseline, Neural CDE | Last-point prediction in the OU setting · OU | MSE on last point: 0.013 ± 0.016 -> 0.018 ± 0.025 | Hurt |
| Use SigLasso instead of the strongest baseline, GRU | Full-path reconstruction in the tumor-growth setting · Tumor growth | L2 error: 0.66 ± 0.09 -> 0.16 ± 0.02 | Helped |
| Use SigLasso instead of the strongest baseline, GRU | Last-point prediction in the tumor-growth setting · Tumor growth | MSE on last point: 2.00 ± 0.38 -> 0.35 ± 0.12 | Helped |
| Use signature-based linear regression (SigLasso) instead of GRU | Training computation, including cross-validation · Experiments | training speed: approximately 10 times faster | Helped |
| Use signature-based linear regression (SigLasso) instead of Neural CDE | Training computation, including cross-validation · Experiments | training speed: 100 times faster | Helped |
Closest papers by meaning
- Zero-shot Imputation with Foundation Inference Models for Dynamical Systems (ICLR 2025 · similarity 0.87)
- Learning Low-dimensional Latent Dynamics from High-dimensional Observations: Non-asymptotics and Lower Bounds (ICML 2024 · similarity 0.87)
- Learning Interpretable Hierarchical Dynamical Systems Models from Time Series Data (ICLR 2025 · similarity 0.87)
- Learning the Causal Structure of Networked Dynamical Systems under Latent Nodes and Structured Noise (AAAI 2024 · similarity 0.85)
- Learning invariant representations of time-homogeneous stochastic dynamical systems (ICLR 2024 · similarity 0.85)
- In-Context Learning of Stochastic Differential Equations with Foundation Inference Models (NeurIPS 2025 · 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.