Demystifying the Neural Tangent Kernel from a Practical Perspective: Can it be trusted for Neural Architecture Search without training?
Jisoo MokByunggook NaJi-Hoon KimDongyoon HanSungroh YoonPublished in: CVPR (2022)
Keyphrases
- neural architecture
- multi layer perceptron
- neural network
- feed forward
- search algorithm
- search space
- basis functions
- multilayer perceptron
- real world
- training samples
- kernel methods
- kernel function
- artificial neural networks
- training set
- input space
- training phase
- training examples
- supervised learning
- pattern recognition
- activation function
- processing units
- learning algorithm