Reinforcement Learning-Based Layer-Wise Quantization For Lightweight Deep Neural Networks.
Juri JungJonghee KimYoungeun KimChangick KimPublished in: ICIP (2020)
Keyphrases
- lightweight
- neural network
- reinforcement learning
- multi layer
- multiple layers
- single layer
- function approximators
- pattern recognition
- function approximation
- model free
- wireless sensor networks
- activation function
- learning classifier systems
- neural nets
- pairwise
- learning algorithm
- neural network model
- learning capabilities
- reinforcement learning algorithms
- multilayer perceptron
- application layer
- back propagation
- state space
- communication infrastructure
- temporal difference
- feed forward
- fuzzy logic
- artificial neural networks
- genetic algorithm
- optimal policy
- dos attacks