Fast Convergence of Natural Gradient Descent for Over-Parameterized Neural Networks.
Guodong ZhangJames MartensRoger B. GrossePublished in: NeurIPS (2019)
Keyphrases
- neural network
- pattern recognition
- weight update
- learning rules
- update rule
- cost function
- fuzzy logic
- back propagation
- real world
- artificial neural networks
- loss function
- fault diagnosis
- multilayer perceptron
- recurrent neural networks
- genetic algorithm
- objective function
- convergence rate
- neural nets
- rule extraction
- lyapunov theory
- fuzzy systems
- real time
- training algorithm
- evolutionary algorithm
- expert systems
- competitive learning
- computer vision
- database