Research map / Time series / Causal and generative time series
STEER : Simple Temporal Regularization For Neural ODE
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 |
|---|---|---|---|
| Add STEER ($b=0.5$) to FFJORD RNODE | FFJORD RNODE · MNIST density estimation | time: 24.37 -> 16.32 | Helped |
| Add STEER ($b=0.5$) to FFJORD RNODE | FFJORD RNODE · CIFAR-10 density estimation | time: 31.84 -> 22.24 | Helped |
| Add STEER ($b=0.5$) to FFJORD RNODE | FFJORD RNODE · MNIST density estimation | bits/dim: 0.973 -> 0.971 | Helped |
| Add STEER to Latent ODE with ODE encoder | Latent ODE (ODE enc) · Human Activity | per-time-point classification accuracy: 0.846 ± 0.013 -> 0.880 ± 0.011 | Helped |
| Add STEER to ODE-RNN | ODE-RNN · Physionet | AUC: 0.833 ± 0.009 -> 0.811 ± 0.007 | Hurt |
| Add STEER to Latent ODE with ODE encoder | Latent ODE (ODE enc) · MuJoCo interpolation, 10 % observed points | MSE (× 10⁻²): 0.360 -> 0.230 | Helped |
| Add STEER to NODE | NODE · MNIST classification | test accuracy: 97.2% ± 0.2 -> 98.3% ± 0.3 | Helped |
| Add STEER to ANODE | ANODE · CIFAR-10 classification | test accuracy: 60.6% ± 0.4 -> 62.1% ± 0.2 | Helped |
| Add STEER ($b=0.124$) | Neural ODE · Stiff ODE, test interval [0,25] | MSE: 3.87 -> 0.52 | Helped |
| Increase STEER interval parameter to $b=0.5$ while setting $t_1+b=t_1^{original}$ (ablation) | FFJORD RNODE · MNIST density estimation | time: 22.32 -> 16.32 | Helped |
| Set $t_1=t_1^{original}$ rather than $t_1+b=t_1^{original}$ (ablation) | FFJORD RNODE + STEER ($b=0.5$) · MNIST density estimation | time: 16.32 -> 25.81 | Hurt |
| Replace uniform end-time sampling with Gaussian sampling (ablation) | STEER on stiff ODE · Stiff ODE | MSE | Mixed |
Closest papers by meaning
- Improving Neural ODE Training with Temporal Adaptive Batch Normalization (NeurIPS 2024 · similarity 0.89)
- How Deep Do We Need: Accelerating Training and Inference of Neural ODEs via Control Perspective (ICML 2024 · similarity 0.89)
- Implicit regularization of deep residual networks towards neural ODEs (ICLR 2024 · similarity 0.88)
- Global Convergence in Neural ODEs: Impact of Activation Functions (ICLR 2025 · similarity 0.87)
- Stable Neural Stochastic Differential Equations in Analyzing Irregular Time Series Data (ICLR 2024 · similarity 0.87)
- Simulation-Free Training of Neural ODEs on Paired Data (NeurIPS 2024 · similarity 0.87)
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.