26. 卷积神经网络 26
LeNet-5 by LeCun in 1998
LeCun, Yann, et al. "Gradient-based learning applied to document
recognition." Proceedings of the IEEE 86.11 (1998): 2278-2324.
57. 卷积运算 57
Zero Pad
CONV layers with stride 1, filters of size FxF,
P = (F-1)/2
有三个参数: F(filter size), S(stride), P(pad)
有时,想由(N - F) / stride + 1得出整数不容易
pad的数量未必四边对称
58. 卷积运算 58
Output size: (N - F) / stride + 1
(224-11)/4+1=54.25 ≠ 25
Alex God was wrong?
Hinton Big God was wrong?
我并不认为两位大神在这里出错,
很有可能他们使用了pad,224→227