【caffe】Layer解读之：Reduction

Layer type: Reduction
头文件位置：./include/caffe/layers/reduction_layer.hpp
CPU 执行源文件位置: ./src/caffe/layers/reduction_layer.cpp
CUDA GPU 执行源文件位置: ./src/caffe/layers/reduction_layer.cu
Reduction层的功能：使用sum或mean等操作作用于输入blob按照给定参数规定的维度。（通俗的讲就是将输入的特征图按照给定的维度进行求和或求平均）。

参数解读

layer {
  name: "rnn_o_concat"
  type: "Concat"
  bottom: "o_1"
  bottom: "o_2"
  bottom: "o_3"
  top: "o"
  concat_param {
    axis: 0
  }
}
layer {
  name: "o_pseudoloss"
  type: "Reduction"
  bottom: "o"
  top: "o_pseudoloss"
  loss_weight: 1
  reduction_param{
       axis: 0//default = 0
       optional ReductionOp operation = SUM //default = SUM
    }
}

根据Reduction的定义，我们知道就是将输入的特征图“o”的第一个维度的特征图都加起来。concat_layer的用法见：【caffe】Layer解读之：Concat

参数定义

参数(ReductionParameter reduction_param)
定义位置 ./src/caffe/proto/caffe.proto:

// Message that stores parameters used by ReductionLayer
message ReductionParameter {
  enum ReductionOp {
    SUM = 1;
    ASUM = 2;
    SUMSQ = 3;
    MEAN = 4;
  }

  optional ReductionOp operation = 1 [default = SUM]; // reduction operation

  // The first axis to reduce to a scalar -- may be negative to index from the
  // end (e.g., -1 for the last axis).
  // (Currently, only reduction along ALL "tail" axes is supported; reduction
  // of axis M through N, where N < num_axes - 1, is unsupported.)
  // Suppose we have an n-axis bottom Blob with shape:
  //     (d0, d1, d2, ..., d(m-1), dm, d(m+1), ..., d(n-1)).
  // If axis == m, the output Blob will have shape
  //     (d0, d1, d2, ..., d(m-1)),
  // and the ReductionOp operation is performed (d0 * d1 * d2 * ... * d(m-1))
  // times, each including (dm * d(m+1) * ... * d(n-1)) individual data.
  // If axis == 0 (the default), the output Blob always has the empty shape
  // (count 1), performing reduction across the entire input --
  // often useful for creating new loss functions.
  optional int32 axis = 2 [default = 0];

  optional float coeff = 3 [default = 1.0]; // coefficient for output
}

番外篇

若想用Reduction解决如下问题：
我想把n*c*h*w的blob变成n*1*h*w（使用SUM），param该如何设定？
答：ReductionLayer 干不了这事儿。因为它只支持从你指定的axis到tail axis为止的reduction操作。它不支持针对某个坐标轴独立做reduction，而是从某个坐标轴开始做到最后一个坐标轴。即无论你的指定哪个坐标轴，它都会默认reduce做到最后一个坐标轴的。n*c*h*w, 你若设axis=1，它就变成n, 若你指定2，它就变成n*c。

正确解决方案（来自网络）：
答：你可以用convolutionLayer做你想做的事儿。如下，注意，kernel_size，num_output，weight的值都必须为1才能做到这个效果。

convolution_param { 
    num_output: 1 
    bias_term: false 
    kernel_size: 1 
    weight_filler { 
        type: 'constant' 
        value: 1 
    } 
}