Neural signature kernels as infinite-width-depth-limits of controlled ResNets.
Nicola Muca CironeMaud LemercierCristopher SalviPublished in: CoRR (2023)
Keyphrases
- network architecture
- depth information
- neural network
- depth map
- kernel function
- kernel methods
- feature space
- signature scheme
- neural model
- neural computation
- neural fuzzy
- artificial neural
- hebbian learning
- signature recognition
- gaussian kernels
- kernel machines
- depth estimation
- multiple kernel learning
- associative memory
- artificial neural networks
- three dimensional