Post-training deep neural network pruning via layer-wise calibration.
Ivan LazarevichAlexander KozlovNikita MalininPublished in: CoRR (2021)
Keyphrases
- neural network
- multi layer
- training algorithm
- feed forward neural networks
- training process
- feedforward neural networks
- training patterns
- back propagation
- neural network training
- multi layer perceptron
- single layer
- backpropagation algorithm
- artificial neural networks
- multiple layers
- search space
- error back propagation
- pairwise
- output layer
- avoid overfitting
- training set
- deep architectures
- neural network model
- multilayer perceptron
- training phase
- adaptive learning rate
- pruning method
- search algorithm
- pattern recognition
- training examples
- fuzzy logic
- online learning
- data sets
- training data
- neural network is trained
- neural network structure
- knn
- hidden layer
- test set
- camera calibration
- restricted boltzmann machine
- eye tracking
- application layer
- camera network