HPPU: An Energy-Efficient Sparse DNN Training Processor with Hybrid Weight Pruning.
Yang WangYubin QinLeibo LiuShaojun WeiShouyi YinPublished in: AICAS (2021)
Keyphrases
- avoid overfitting
- training process
- parallel processing
- training set
- high dimensional
- search space
- wireless sensor networks
- training data
- sensor networks
- supervised learning
- data sets
- multi core processors
- compressive sensing
- training algorithm
- training examples
- high speed
- real time
- sparse coding
- online learning
- single chip
- support vector machine
- pruning method