通道随机混合操作（Channel Shuffle Operation）可以看成“重塑-转置-重塑”（“reshape-transpose-
    reshape”）操作。这里假设把4个Feature Maps级联后，共1024个Channels。现在把这1024个Channels随机打乱混合。首先把Channels重塑为（g, c），其中 g 表示分组数目，c=1024/g。然后把它转置一下为（c, g）。然后把它重塑为1024个通道。

    2. def channel_shuffle(x, groups):
    3.     batchsize, num_channels, height, width = x.data.size()
    5.     channels_per_group = num_channels // groups
    7.     # reshape
    8.     x = x.view(batchsize, groups,
    9.                channels_per_group, height, width)
    11.     # transpose
    12.     # - contiguous() required if transpose() is used before view().
    13.     #   See https://github.com/pytorch/pytorch/issues/764
    14.     x = torch.transpose(x, 1, 2).contiguous()
    16.     # flatten