Implicit Bias of Gradient Descent for Wide Two-layer Neural Networks Trained with the Logistic Loss.
Lénaïc ChizatFrancis R. BachPublished in: COLT (2020)
Keyphrases
- neural network
- multilayer perceptron
- multi layer
- trained neural networks
- multi layer perceptron
- training process
- trained neural network
- single layer
- auto associative
- synaptic weights
- artificial neural networks
- pattern recognition
- learning rules
- back propagation
- backpropagation algorithm
- cost function
- fuzzy systems
- rule extraction
- neural network model
- training set
- objective function
- neural nets
- loss function
- neural learning
- activation function
- fuzzy logic
- multilayer neural network
- wide range
- training algorithm
- genetic algorithm
- feed forward
- back propagation neural network
- elman network
- neural network is trained
- feedforward neural networks
- feed forward neural networks
- feature space
- machine learning