Channel Pruning for Deep Neural Networks Via a Relaxed Groupwise Splitting Method.
Biao YangJiancheng LyuShuai ZhangYingyong QiJack XinPublished in: AI4I (2019)
Keyphrases
- neural network
- high precision
- high accuracy
- cost function
- dynamic programming
- similarity measure
- segmentation method
- significant improvement
- computational complexity
- objective function
- k means
- preprocessing
- experimental evaluation
- computational cost
- support vector machine
- input data
- fully automatic
- pairwise
- feature space
- energy function
- training data
- neural nets
- pruning method