Impact of Mixed Precision Techniques on Training and Inference Efficiency of Deep Neural Networks.
Marion DörrichMingcheng FanAndreas M. KistPublished in: IEEE Access (2023)
Keyphrases
- neural network
- training process
- training algorithm
- feedforward neural networks
- feed forward neural networks
- backpropagation algorithm
- pattern recognition
- structured prediction
- multi layer perceptron
- computational complexity
- supervised learning
- back propagation
- genetic algorithm
- error back propagation
- multi layer
- high precision
- recurrent neural networks
- deep architectures
- radial basis function network
- neural network model
- hidden layer
- test set
- neural network training
- training samples
- precision and recall
- fuzzy logic
- training set
- recurrent networks
- activation function
- training data
- neural network structure
- decision trees
- deep learning
- expert systems
- training phase
- probabilistic inference
- neural nets
- multilayer perceptron