caffe相关--Layers

http://caffe.berkeleyvision.org/tutorial/layers.html

Google Protocol Buffers 简介（一）：http://www.jianshu.com/p/7de98349cadd
Data Layers
Data enters Caffe through data layers: they lie at the bottom of nets.
Note that the Python Layer can be useful for create custom data layers.

import caffe
import numpy as np


class EuclideanLossLayer(caffe.Layer):
    """
    Compute the Euclidean Loss in the same manner as the C++ EuclideanLossLayer
    to demonstrate the class interface for developing layers in Python.
    """

    def setup(self, bottom, top):
        # check input pair
        if len(bottom) != 2:
            raise Exception("Need two inputs to compute distance.")

    def reshape(self, bottom, top):
        # check input dimensions match
        if bottom[0].count != bottom[1].count:
            raise Exception("Inputs must have the same dimension.")
        # difference is shape of inputs
        self.diff = np.zeros_like(bottom[0].data, dtype=np.float32)
        # loss output is scalar
        top[0].reshape(1)

    def forward(self, bottom, top):
        self.diff[...] = bottom[0].data - bottom[1].data
        top[0].data[...] = np.sum(self.diff**2) / bottom[0].num / 2.

    def backward(self, top, propagate_down, bottom):
        for i in range(2):
            if not propagate_down[i]:
                continue
            if i == 0:
                sign = 1
            else:
                sign = -1
            bottom[i].diff[...] = sign * self.diff / bottom[i].num

propagate_down： const vector< bool > &
a vector with equal length to bottom, with each index indicating whether to propagate the error gradients down to the bottom blob at the corresponding index

http://www.jianshu.com/p/622b8bb24589

reshape函数中里面必须把top的shape给定下来，比如说，这是一个loss layer，那么这个层的输出就是一个loss值，所以这个top的shape就是1*1，所以就这样写：

top[0].reshape(1,1)

因为这里比较简单，就不啰嗦了，只是比较奇怪的是，为什么反馈的时候会对标签y做偏导计算呢？好奇怪！！当然如果不要将y理解成标签，将这种欧氏loss理解层和上面介绍的loss同样的效果，那么对y求偏导也是可以理解的。

Vision Layers
Vision layers usually take images as input and produce other images as output, although they can take data of other types and dimensions.

Convolution Layer
Input：
n * c_i * h_i * w_i
Output：
n * c_o * h_o * w_o, where h_o = (h_i + 2 * pad_h - kernel_h) / stride_h + 1 and w_o likewise.

学习率 权值衰减相关
http://yufeigan.github.io/2014/11/29/Deep-Learning-%E4%BC%98%E5%8C%96%E6%96%B9%E6%B3%95%E6%80%BB%E7%BB%93/

权值初始化：http://www.jianshu.com/p/03009cfdf733

CPU implementation: ./src/caffe/layers/conv_layer.cpp：
https://github.com/BVLC/caffe/blob/master/src/caffe/layers/conv_layer.cpp#L61

this指针：
this 是 C++ 中的一个关键字，也是一个 const 指针，它指向当前对象，通过它可以访问当前对象的所有成员。

所谓当前对象，是指正在使用的对象。例如对于stu.show();，stu 就是当前对象，this 就指向 stu。

Blob< int > kernel_shape_
The spatial dimensions of a filter kernel.

http://caffe.berkeleyvision.org/doxygen/classcaffe_1_1BaseConvolutionLayer.html