TNPU: an efficient accelerator architecture for training convolutional neural networks.
Jiajun LiGuihai YanWenyan LuShuhao JiangShijun GongJingya WuJunchao YanXiaowei LiPublished in: ASP-DAC (2019)
Keyphrases
- convolutional neural networks
- real time
- management system
- convolutional network
- architectural design
- training phase
- feedforward artificial neural networks
- online learning
- training examples
- test set
- data sets
- layered architecture
- distributed architecture
- parallel implementation
- training set
- machine learning
- training algorithm
- training process
- software architecture
- text classification
- semi supervised
- neural network