最近在打基础,大致都和向量有关,从比较基础的人工智能常用算法开始,以下是对BP算法研究的一个小节。
本文只是自我思路的整理,其中举了个例子,已经对一些难懂的地方做了解释,有兴趣恰好学到人工智能对这块不能深入理解的,可以参考本文。
通过带*的权重值重新计算误差,发现误差为0.18,比老误差0.19小,则继续迭代,得神经元的计算结果更加逼近目标值0.5
感想
在一个复杂样本空间下,对输入和输出进行拟合
(1) 多少个hidden unit才能符合需要(hidden unit就是图中的P,Q)
(2) 多少层unit才能符合需要(本例为1层)
(3) 如果有n层,每层m个unit,k个输入,1个输出,那么就有K*m^(n+1)条边,每条边有一个权重值,这个计算量非常巨大
(4) 如果k个输入,1个输出,相当于将k维空间,投射到一个1维空间,是否可以提供足够的准确性,如果是k个输入,j个输出,j比k大,是否是一个升维的过程,是否有价值?
收获
1) 了解偏导。
2) 了解梯度。
3) 产生新的思考
参考文献:
1 http://en.wikipedia.org/wiki/Linear_least_squares_(mathematics)
2 http://www.rgu.ac.uk/files/chapter3%20-%20bp.pdf
3 http://www.cedar.buffalo.edu/~srihari/CSE574/Chap5/Chap5.3-BackProp.pdf
验证代码:
#include "stdio.h"
#include <math.h>
const double e = 2.7182818;
int main(void)
{
double input[] = {0.35,0.9};
double matrix_1[2][2]={
{0.1,0.4},
{0.8,0.6},
};
double matrix_2[] = { 0.3,0.9 };
for(int s= 0; s<1000; ++s)
{
double tmp[] = {0.0,0.0};
double value = 0.0;
{
for(int i = 0;i<2;++i)
{
for(int j = 0;j<2;++j)
{
tmp[i] += input[j]*matrix_1[j][i];
}
tmp[i] = 1/(1+pow(e,-1*tmp[i]));
}
for(int i = 0;i<2;++i)
{
value += tmp[i]*matrix_2[i];
}
value = 1/(1+pow(e,-1*value));
}
double RMSS = (0.5)*( value - 0.5)*(value-0.5);
printf("%f,%f\n",value,RMSS);
if(value - 0.5 < 0.01)
{
break;
}
double E = value - 0.5;
matrix_2[0] = matrix_2[0] - E*value*(1-value)*tmp[0];
matrix_2[1] = matrix_2[1] - E*value*(1-value)*tmp[1];
//printf("##%f,%f\n",matrix_2[0],matrix_2[1]);
matrix_1[0][0] = matrix_1[0][0] - E*value*(1-value)*matrix_2[0]*tmp[0]*(1-tmp[0])*matrix_1[0][0];
matrix_1[1][0] = matrix_1[1][0] - E*value*(1-value)*matrix_2[0]*tmp[0]*(1-tmp[0])*matrix_1[1][0];
matrix_1[0][1] = matrix_1[0][1] - E*value*(1-value)*matrix_2[0]*tmp[0]*(1-tmp[1])*matrix_1[0][1];
matrix_1[1][1] = matrix_1[1][1] - E*value*(1-value)*matrix_2[0]*tmp[0]*(1-tmp[1])*matrix_1[1][1];
//printf("##%f,%f\n",matrix_1[0][0],matrix_1[1][0]);
//printf("##%f,%f\n",matrix_1[0][1],matrix_1[1][1]);
}
return 0;
}
给出一个带有”增加充量项“BPANN。
有两个输入单元,两个隐藏单元,一个输出单元,三层
- # Back-Propagation Neural Networks
- #
- import math
- import random
- import string
- random.seed(0)
- # calculate a random number where: a <= rand < b
- def rand(a, b):
- return (b-a)*random.random() + a
- # Make a matrix (we could use NumPy to speed this up)
- def makeMatrix(I, J, fill=0.0):
- m = []
- for i in range(I):
- m.append([fill]*J)
- return m
- # our sigmoid function, tanh is a little nicer than the standard 1/(1+e^-x)
- #使用双正切函数代替logistic函数
- def sigmoid(x):
- return math.tanh(x)
- # derivative of our sigmoid function, in terms of the output (i.e. y)
- # 双正切函数的导数,在求取输出层和隐藏侧的误差项的时候会用到
- def dsigmoid(y):
- return 1.0 - y**2
- class NN:
- def __init__(self, ni, nh, no):
- # number of input, hidden, and output nodes
- # 输入层,隐藏层,输出层的数量,三层网络
- self.ni = ni + 1 # +1 for bias node
- self.nh = nh
- self.no = no
- # activations for nodes
- self.ai = [1.0]*self.ni
- self.ah = [1.0]*self.nh
- self.ao = [1.0]*self.no
- # create weights
- #生成权重矩阵,每一个输入层节点和隐藏层节点都连接
- #每一个隐藏层节点和输出层节点链接
- #大小:self.ni*self.nh
- self.wi = makeMatrix(self.ni, self.nh)
- #大小:self.ni*self.nh
- self.wo = makeMatrix(self.nh, self.no)
- # set them to random vaules
- #生成权重,在-0.2-0.2之间
- for i in range(self.ni):
- for j in range(self.nh):
- self.wi[i][j] = rand(-0.2, 0.2)
- for j in range(self.nh):
- for k in range(self.no):
- self.wo[j][k] = rand(-2.0, 2.0)
- # last change in weights for momentum
- #?
- self.ci = makeMatrix(self.ni, self.nh)
- self.co = makeMatrix(self.nh, self.no)
- def update(self, inputs):
- if len(inputs) != self.ni-1:
- raise ValueError('wrong number of inputs')
- # input activations
- # 输入的激活函数,就是y=x;
- for i in range(self.ni-1):
- #self.ai[i] = sigmoid(inputs[i])
- self.ai[i] = inputs[i]
- # hidden activations
- #隐藏层的激活函数,求和然后使用压缩函数
- for j in range(self.nh):
- sum = 0.0
- for i in range(self.ni):
- #sum就是《ml》书中的net
- sum = sum + self.ai[i] * self.wi[i][j]
- self.ah[j] = sigmoid(sum)
- # output activations
- #输出的激活函数
- for k in range(self.no):
- sum = 0.0
- for j in range(self.nh):
- sum = sum + self.ah[j] * self.wo[j][k]
- self.ao[k] = sigmoid(sum)
- return self.ao[:]
- #反向传播算法 targets是样本的正确的输出
- def backPropagate(self, targets, N, M):
- if len(targets) != self.no:
- raise ValueError('wrong number of target values')
- # calculate error terms for output
- #计算输出层的误差项
- output_deltas = [0.0] * self.no
- for k in range(self.no):
- #计算k-o
- error = targets[k]-self.ao[k]
- #计算书中公式4.14
- output_deltas[k] = dsigmoid(self.ao[k]) * error
- # calculate error terms for hidden
- #计算隐藏层的误差项,使用《ml》书中的公式4.15
- hidden_deltas = [0.0] * self.nh
- for j in range(self.nh):
- error = 0.0
- for k in range(self.no):
- error = error + output_deltas[k]*self.wo[j][k]
- hidden_deltas[j] = dsigmoid(self.ah[j]) * error
- # update output weights
- # 更新输出层的权重参数
- # 这里可以看出,本例使用的是带有“增加冲量项”的BPANN
- # 其中,N为学习速率 M为充量项的参数 self.co为冲量项
- # N: learning rate
- # M: momentum factor
- for j in range(self.nh):
- for k in range(self.no):
- change = output_deltas[k]*self.ah[j]
- self.wo[j][k] = self.wo[j][k] + N*change + M*self.co[j][k]
- self.co[j][k] = change
- #print N*change, M*self.co[j][k]
- # update input weights
- #更新输入项的权重参数
- for i in range(self.ni):
- for j in range(self.nh):
- change = hidden_deltas[j]*self.ai[i]
- self.wi[i][j] = self.wi[i][j] + N*change + M*self.ci[i][j]
- self.ci[i][j] = change
- # calculate error
- #计算E(w)
- error = 0.0
- for k in range(len(targets)):
- error = error + 0.5*(targets[k]-self.ao[k])**2
- return error
- #测试函数,用于测试训练效果
- def test(self, patterns):
- for p in patterns:
- print(p[0], '->', self.update(p[0]))
- def weights(self):
- print('Input weights:')
- for i in range(self.ni):
- print(self.wi[i])
- print()
- print('Output weights:')
- for j in range(self.nh):
- print(self.wo[j])
- def train(self, patterns, iterations=1000, N=0.5, M=0.1):
- # N: learning rate
- # M: momentum factor
- for i in range(iterations):
- error = 0.0
- for p in patterns:
- inputs = p[0]
- targets = p[1]
- self.update(inputs)
- error = error + self.backPropagate(targets, N, M)
- if i % 100 == 0:
- print('error %-.5f' % error)
- def demo():
- # Teach network XOR function
- pat = [
- [[0,0], [0]],
- [[0,1], [1]],
- [[1,0], [1]],
- [[1,1], [0]]
- ]
- # create a network with two input, two hidden, and one output nodes
- n = NN(2, 2, 1)
- # train it with some patterns
- n.train(pat)
- # test it
- n.test(pat)
- if __name__ == '__main__':
- demo()
>>> ================================ RESTART ================================
>>>
error 0.94250
error 0.04287
error 0.00348
error 0.00164
error 0.00106
error 0.00078
error 0.00125
error 0.00053
error 0.00044
error 0.00038
([0, 0], '->', [0.03668584043139609])
([0, 1], '->', [0.9816625517128087])
([1, 0], '->', [0.9815264813097478])
([1, 1], '->', [-0.03146072993485337])
>>>