A type of generalization error induced by initialization in deep neural networks.
Yaoyu ZhangZhi-Qin John XuTao LuoZheng MaPublished in: CoRR (2019)
Keyphrases
- generalization error
- neural network
- model selection
- cross validation
- learning algorithm
- active learning
- upper bound
- binary classification
- training data
- training set
- target function
- sample complexity
- training error
- classification error
- sample size
- generalization error bounds
- supervised learning
- linear classifiers
- subspace information criterion
- back propagation
- fuzzy logic
- learning machines
- artificial neural networks
- conditional expectation
- recurrent neural networks
- boosting algorithms
- computational complexity
- training set size
- support vector
- perceptron learning
- genetic algorithm