DrowsyNet: Convolutional neural networks with runtime power-accuracy tunability using inference-stage dropout.
Ren-Shuo LiuYun-Chen LoYuan-Chun LuoChih-Yu ShenCheng-Ju LeePublished in: VLSI-DAT (2018)
Keyphrases
- convolutional neural networks
- high accuracy
- classification accuracy
- database
- power consumption
- prediction accuracy
- convolutional network
- improved accuracy
- dynamic bayesian networks
- inference engine
- belief networks
- computational efficiency
- error rate
- computational cost
- computational complexity
- website
- social networks