Benign overfitting of non-smooth neural networks beyond lazy training.
Xingyu XuYuantao GuPublished in: AISTATS (2023)
Keyphrases
- neural network
- training process
- training algorithm
- feed forward neural networks
- feedforward neural networks
- neural network training
- cross validation
- pattern recognition
- back propagation
- multi layer perceptron
- training set
- training error
- radial basis function network
- training phase
- backpropagation algorithm
- test set
- recurrent networks
- avoid overfitting
- recurrent neural networks
- fuzzy logic
- error back propagation
- serious games
- activation function
- multi layer
- neural nets
- feed forward
- training samples
- supervised learning
- artificial neural networks
- hidden layer
- fuzzy systems
- associative memory
- neural network model
- self organizing maps
- training patterns
- fault diagnosis
- training examples
- averaged one dependence estimators