Caffe入门：caffe中的Power层

Power：数据的缩放操作

类型： Power

layer {
  name: "layer"
  bottom: "in"
  top: "out"
  type: "Power"
  power_param {
    power: 1
    scale: 1
    shift: 0
  }
}

可选参数：

power [default 1]

scale [default 1]

shift [default 0]

POWER层通过 (shift + scale * x) ^ power计算每一个输入x的输出。

实际应用：

# H-swish 激活函数(MobileNet_V3)
layer {
  name: "conv1/shift"
  type: "Power"
  bottom: "conv1"
  top: "conv1/shift"
  power_param {
    power: 1
    scale: 1
    shift: 3
  }
}
layer {
  name: "conv1/shift-relu"
  type: "ReLU6"
  bottom: "conv1/shift"
  top: "conv1/shift"
}
layer {
  name: "conv1/shift-div"
  type: "Power"
  bottom: "conv1/shift"
  top: "conv1/shift-div"
  power_param {
    power: 1
    scale: 0.1666666667
    shift: 0
  }
}
layer {
  name: "conv1/hswish"
  type: "Eltwise"
  bottom: "conv1"
  bottom: "conv1/shift-div"
  top: "conv1/hswish"
  eltwise_param {
    operation: PROD  #点乘
  }
}