Framing RNN as a kernel method: A neural ODE approach.
Adeline FermanianPierre MarionJean-Philippe VertGérard BiauPublished in: CoRR (2021)
Keyphrases
- kernel methods
- recurrent neural networks
- neural model
- learning problems
- neural network
- nearest neighbor
- network architecture
- support vector
- kernel function
- support vector machine
- machine learning
- kernel matrix
- learning tasks
- feature space
- ordinary differential equations
- reproducing kernel hilbert space
- reproducing kernel
- differential equations
- artificial neural networks
- multiscale
- learning algorithm