AC/DC: Alternating Compressed/DeCompressed Training of Deep Neural Networks.
Alexandra PesteEugenia IofinovaAdrian VladuDan AlistarhPublished in: CoRR (2021)
Keyphrases
- neural network
- data compression
- training process
- training algorithm
- compressed data
- feed forward neural networks
- feedforward neural networks
- multi layer perceptron
- neural network training
- back propagation
- deep architectures
- neural nets
- data structure
- training phase
- file size
- pattern recognition
- compression ratio
- feed forward
- neural network structure
- fuzzy logic
- training set
- error back propagation
- multi layer
- backpropagation algorithm
- multilayer perceptron
- compression algorithm
- activation function
- training examples
- supervised learning
- artificial neural networks