Gradient Descent Finds Global Minima of Deep Neural Networks.
Simon S. DuJason D. LeeHaochuan LiLiwei WangXiyu ZhaiPublished in: CoRR (2018)
Keyphrases
- global minima
- neural network
- global minimum
- cost function
- learning rules
- pattern recognition
- artificial neural networks
- energy minimization
- error function
- fuzzy logic
- multilayer perceptron
- back propagation
- objective function
- training process
- recurrent neural networks
- neural network model
- genetic algorithm
- feed forward
- self organizing maps
- energy function
- loss function
- multi view
- simulated annealing
- dynamic programming