深度學總結：weight_initialization

weight_initialization Occam’s razor：簡單一刀切設置0或者1 General rule for setting weights The general rule for setting the weights in a neural network is to set them to be close to zero without being too small

>>阅读原文<<