Learning Rate Dropout

Huangxing Lin, Weihong Zeng, Xinghao Ding +3

The performance of a deep neural network is highly dependent on its training, and finding better local optimal solutions is the goal of many optimization algorithms. However, exist…