Artificial Intelligence
神经常微分方程
Neural Ordinary Differential EquationsA class of continuous-depth neural networks that model the derivative of the hidden state as a function of the hidden state and input, offering a continuous-time perspective on deep learning architectures. / 一类连续深度神经网络，将隐藏状态的导数建模为隐藏状态和输入的函数，为深度学习架构提供了连续时间的视角。
- Knowledge type
- 数学理论
- First proposed or discovered
- 2018
- Proposer
- Chen, Ricky T. Q.; Rubanova, Yulia; Bettencourt, Jesse; Duvenaud, David K.
- Creator
- 农场主
- Last updated
- 2026-11-09 16:06:31
Sub-knowledge points
Continuous-depth architectureImplicit layersNeural ODE solvers
Higher-order knowledge points
Deep Learning / 工艺技术Differential Equations
Main parameters and performance indicators
f: The neural network function mapping state and time to derivativet: The time variablex: The input data
Open problems
No unresolved problem recorded yet
Related literature
ranked by community starsHypernetwork-based policies for neural-ordinary-differential-equation controlOpen source
Related products and links
No item recorded yet
Creation and modification trail
1 recordsRecord title: Neural Ordinary Differential Equations.
Node update feedback
review required