用于图像分类的卷积神经网络中激活函数的设计
王红霞,周家奇,辜承昊,林泓
Design of activation function in CNN for image classification
Hong-xia WANG,Jia-qi ZHOU,Cheng-hao GU,Hong LIN
表 3
CIFAR-100上的实验结果
Tab.3
Results on CIFAR-100
激活函数
loss
t
ACC
t
loss
v
ACC
v
relu
0.748 9
0.770 2
1.706 7
0.595 2
softsign
1.041 3
0.699 0
1.607 4
0.585 0
relu-softplus
0.853 8
0.749 4
1.596 2
0.603 1
relu-softsign
0.743 1
0.778 2
1.582 3
0.614 3