High-Efficiency Convolutional Ternary Neural Networks with Custom Adder Trees and Weight Compression.
Adrien Prost-BoucleAlban BourgeFrédéric PétrotPublished in: ACM Trans. Reconfigurable Technol. Syst. (2018)
Keyphrases
- high efficiency
- neural network
- network architecture
- high accuracy
- wavelet tree
- real and synthetic datasets
- compression algorithm
- pattern recognition
- data compression
- decision trees
- domain specific
- weight update
- fuzzy logic
- compression scheme
- back propagation
- fuzzy decision trees
- image compression
- density based clustering
- deep learning
- neural network model
- compression ratio
- self organizing maps
- artificial neural networks
- data structure
- tree structure
- recurrent neural networks
- multi dimensional
- arbitrary shape
- memory space
- synaptic weights
- feed forward
- data flow
- multilayer perceptron
- fault diagnosis
- unsupervised learning
- wavelet transform
- genetic algorithm