Research map / Time series / Causal and generative time series
Modeling Irregular Time Series with Continuous Recurrent Units
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 CRU instead of RKN-Δt | Irregular time-series interpolation · USHCN | MSE (×10⁻²): 0.009 ± 0.002 -> 0.016 ± 0.006 | Hurt |
| Use CRU instead of RKN-Δt | Irregular time-series interpolation · Physionet | MSE (×10⁻²): 0.186 ± 0.030 -> 0.182 ± 0.091 | Helped |
| Use CRU instead of RKN | Irregular time-series extrapolation · USHCN | MSE (×10⁻²): 1.478 ± 0.283 -> 1.273 ± 0.066 | Helped |
| Use CRU instead of mTAND | Irregular time-series extrapolation · Physionet | MSE (×10⁻²): 0.340 ± 0.020 -> 0.629 ± 0.093 | Hurt |
| Use CRU instead of RKN-Δt | Pendulum image interpolation · Pendulum | MSE (×10⁻³): 1.903 ± 0.137 -> 0.996 ± 0.052 | Helped |
| Use CRU instead of RKN-Δt | Pendulum angle prediction from noisy images · Pendulum regression | MSE (×10⁻³): 5.092 ± 0.395 -> 4.626 ± 1.072 | Helped |
| Use CRU instead of GRU-Δt | Pendulum angle prediction with uncertainty estimates · Pendulum regression | Gaussian NLL: -5.45 ± 0.09 -> -5.49 ± 0.05 | Helped |
| Use FCRU instead of CRU (ablation) | Irregular time-series interpolation · USHCN | MSE (×10⁻²): 0.016 ± 0.006 -> 0.013 ± 0.004 | Hurt |
| Use FCRU instead of CRU (ablation) | Irregular time-series interpolation · Physionet | MSE (×10⁻²): 0.182 ± 0.091 -> 0.194 ± 0.007 | Hurt |
| Use FCRU instead of CRU (ablation) | Irregular time-series extrapolation · USHCN | MSE (×10⁻²): 1.273 ± 0.066 -> 1.569 ± 0.321 | Hurt |
| Use FCRU instead of CRU (ablation) | Irregular time-series extrapolation · Physionet | MSE (×10⁻²): 0.629 ± 0.093 -> 0.714 ± 0.036 | Hurt |
| Use FCRU instead of CRU (ablation) | Pendulum image interpolation · Pendulum | MSE (×10⁻³): 0.996 ± 0.052 -> 1.386 ± 0.162 | Hurt |
Closest papers by meaning
- Latent Ordinary Differential Equations for Irregularly-Sampled Time Series (NeurIPS 2019 · similarity 0.88)
- Legendre Memory Units: Continuous-Time Representation in Recurrent Neural Networks (NeurIPS 2019 · similarity 0.87)
- GRU-ODE-Bayes: Continuous Modeling of Sporadically-Observed Time Series (NeurIPS 2019 · similarity 0.87)
- ContiFormer: Continuous-Time Transformer for Irregular Time Series Modeling (NeurIPS 2023 · similarity 0.87)
- Transformation of ReLU-based recurrent neural networks from discrete-time to continuous-time (ICML 2020 · similarity 0.86)
- Parallelizing non-linear sequential models over the sequence length (ICLR 2024 · similarity 0.84)
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.