Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Dissecting Neural ODEs
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 input-layer augmentation instead of no augmentation | Neural ODE · MNIST | Test Acc.: 96.8 -> 99.1 | Helped |
| Use input-layer augmentation instead of 0-augmentation | Neural ODE · CIFAR | Test Acc.: 70.8 -> 73.4 | Helped |
| Use input-layer augmentation instead of no augmentation | Neural ODE · CIFAR | NFE: 93 -> 65 | Helped |
| Use input-layer augmentation instead of 0-augmentation | Neural ODE · CIFAR | NFE: 169 -> 65 | Helped |
| Use second-order dynamics instead of no augmentation | Neural ODE · MNIST | Param.[K]: 21.4 -> 20.0 | Helped |
| Use second-order dynamics instead of no augmentation | Neural ODE · CIFAR | Param.[K]: 37.1 -> 34.6 | Helped |
| Use second-order dynamics instead of input-layer augmentation | Neural ODE · CIFAR | Test Acc.: 73.4 -> 72.8 | Hurt |
| Use second-order dynamics instead of input-layer augmentation | Neural ODE · MNIST | NFE: 44 -> 43 | Helped |
| Use second-order dynamics instead of input-layer augmentation | Neural ODE · CIFAR | NFE: 65 -> 59 | Helped |
| Use tanh or ELU rather than sigmoid, ReLU, or softplus as the final vector-field activation | Neural ODE on half-moons · half-moons | classification performance | Helped |
Closest papers by meaning
- Neural Differential Equations for Learning to Program Neural Nets Through Continuous Learning Rules (NeurIPS 2022 · similarity 0.91)
- Neural Ordinary Differential Equations (NeurIPS 2018 · similarity 0.91)
- How Deep Do We Need: Accelerating Training and Inference of Neural ODEs via Control Perspective (ICML 2024 · similarity 0.89)
- Global Convergence in Neural ODEs: Impact of Activation Functions (ICLR 2025 · similarity 0.87)
- Neural ODE Transformers: Analyzing Internal Dynamics and Adaptive Fine-tuning (ICLR 2025 · similarity 0.87)
- ControlSynth Neural ODEs: Modeling Dynamical Systems with Guaranteed Convergence (NeurIPS 2024 · similarity 0.86)
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.