Symplectic Adjoint Method for Exact Gradient of Neural ODE with Minimal Memory.
Takashi MatsubaraYuto MiyatakeTakaharu YaguchiPublished in: NeurIPS (2021)
Keyphrases
- preprocessing
- cost function
- dynamic programming
- detection method
- neural network
- prior knowledge
- memory requirements
- experimental evaluation
- computational cost
- synthetic data
- support vector machine svm
- memory usage
- optimal control
- high precision
- edge detection
- support vector machine
- significant improvement
- computational complexity
- image processing