Neural signature kernels as infinite-width-depth-limits of controlled ResNets.
Nicola Muca CironeMaud LemercierCristopher SalviPublished in: ICML (2023)
Keyphrases
- neural network
- network architecture
- depth map
- kernel function
- depth information
- kernel methods
- data sets
- support vector
- feature space
- linear combination
- neural computation
- neural fuzzy
- signature recognition
- multiple kernel learning
- nonlinear predictive control
- signature scheme
- bio inspired
- neural model
- positive definite
- biologically plausible
- associative memory
- machine learning