Controlling the Inductive Bias of Wide Neural Networks by Modifying the Kernel's Spectrum.
Amnon GeifmanDaniel BarzilaiRonen BasriMeirav GalunPublished in: Trans. Mach. Learn. Res. (2024)
Keyphrases
- inductive bias
- neural network
- inductive learning
- machine learning
- pac learning
- kernel methods
- prior knowledge
- hypothesis space
- domain knowledge
- training examples
- kernel function
- support vector
- feature space
- domain theory
- supervised learning
- knowledge acquisition
- small number
- learning process
- training set
- reinforcement learning
- data mining