DiracNets: Training Very Deep Neural Networks Without Skip-Connections.
Sergey ZagoruykoNikos KomodakisPublished in: CoRR (2017)
Keyphrases
- neural network
- training process
- training algorithm
- feedforward neural networks
- multi layer perceptron
- deep architectures
- feed forward neural networks
- neural network training
- genetic algorithm
- backpropagation algorithm
- test set
- back propagation
- artificial neural networks
- training patterns
- recurrent networks
- neural nets
- fuzzy logic
- activation function
- pattern recognition
- training phase
- error back propagation
- radial basis function network
- multilayer perceptron
- fault diagnosis
- training samples
- hidden layer
- real time
- fuzzy systems
- recurrent neural networks
- deep learning
- online learning
- training speed
- adaptive resonance theory
- semi supervised
- website
- machine learning