Tensor Programs IIb: Architectural Universality Of Neural Tangent Kernel Training Dynamics.
Greg YangEtai LittwinPublished in: ICML (2021)
Keyphrases
- recurrent networks
- neural network
- support vector
- training set
- high level
- network architecture
- training process
- recurrent neural networks
- kernel function
- cellular automata
- kernel methods
- biologically inspired
- training algorithm
- supervised learning
- linear svm
- artificial neural networks
- training examples
- dynamical systems
- high order
- software architecture
- feed forward
- online learning
- dynamic model
- dimensionality reduction
- computer software