LeNet&AlexNet

AlexNet

第一层

AlexNet	LeNet
image(3X224X224)	image(32X32)
11X11 Conv(96), stride 4	5X5 Conv,PAD2
3X3 MaxPool, stride 2	2x2 AvrgPool

第二层

AlexNet	LeNet
5X5 Conv(256),pad 2
3X3 MaxPooling,stride 2
3X3 Conv, pad 1
3X3 Conv, pad 1
3X3 Conv, pad 1
3X3 MaxPool, stride 2

第三层

AlexNet	LeNet
Dense(4096)	Dense(84)
Dense(4096)	Dense(84)
Dense(1000)	Dense(10)

更多细节

• 激活函数从Sigmoid变到了ReLu(减缓梯度消失)
• 隐藏全连接层后加入了丢弃层
• 数据增加

\[ \text{out}(N_i, C_{\text{out}_j}) = \text{bias}(C_{\text{out}_j}) + \sum_{k = 0}^{C_{\text{in}} - 1} \text{weight}(C_{\text{out}_j}, k) \star \text{input}(N_i, k) \]