T1000: Mitigating the memory footprint of convolution neural networks with decomposition and re-fusion.
Changxi LiuHailong YangRui WangZhongzhi LuanDepei QianPublished in: Future Gener. Comput. Syst. (2018)
Keyphrases
- memory footprint
- neural network
- memory usage
- information fusion
- pattern recognition
- data fusion
- multi sensor
- decomposition methods
- fuzzy logic
- back propagation
- genetic algorithm
- decomposition method
- fusion method
- network architecture
- self organizing maps
- multiscale
- training process
- fusion algorithm
- infrared
- multilayer perceptron
- fusion framework
- decomposition algorithm
- sensor fusion
- convolution kernel
- mesh connected
- fusion model
- image decomposition
- artificial neural networks
- fourier domain
- fault diagnosis
- image reconstruction
- feed forward
- image fusion