Deep learning versus kernel learning: an empirical study of loss landscape geometry and the time evolution of the Neural Tangent Kernel.
Stanislav FortGintare Karolina DziugaiteMansheej PaulSepideh KharaghaniDaniel M. RoySurya GanguliPublished in: CoRR (2020)
Keyphrases
- kernel learning
- deep learning
- kernel function
- multiple kernel learning
- kernel matrix
- unsupervised learning
- machine learning
- semi supervised
- kernel methods
- neural network
- weakly supervised
- support vector regression
- learning tasks
- mental models
- optimal kernel
- dimensionality reduction
- distance metric
- support vector machine
- model selection
- feature space
- low rank
- metric learning
- object recognition
- decision trees
- feature selection
- higher order
- multi class