Neural Tangent Kernel Beyond the Infinite-Width Limit: Effects of Depth and Initialization.
Mariia SeleznovaGitta KutyniokPublished in: ICML (2022)
Keyphrases
- network architecture
- neural network
- support vector
- feature space
- bio inspired
- kernel space
- biologically plausible
- kernel function
- associative memory
- fixed width
- kernel regression
- infinite dimensional
- component analysis
- biologically inspired
- depth map
- similarity measure
- three dimensional
- image segmentation
- clustering algorithm