原理参考：PLA算法
python实现：
import numpy as np
#创建数据集
def createdata():
x = np.array([[3,-3],[4,-3],[1,1],[1,2]])
y = np.array([-1, -1, 1, 1])
return x, y
#感知机模型
class Perceptron:
def __init__(self,x,y):
self.x=x
self.y=y
self.w=np.zeros(self.x.shape[1]) #初始化权重w
self.b = 0
def sign(self, w, b, x):
y = np.dot(x, w)+b
return int(y)
def update(self,label_i,data_i):
tmp = label_i*data_i
self.w += tmp
self.b += label_i
def train(self):
isFind=False
while not isFind:
count=0
for i in range(self.x.shape[0]):
tmp_y=self.sign(self.w, self.b, self.x[i,:])
if tmp_y*self.y[i]<=0: #如果误分类
count+=1
self.update(self.y[i],self.x[i,:])
if count==0:
print('最终训练得到的w和b为：', self.w, self.b)
isFind=True
return self.w, self.b
if __name__ == '__main__':
x, y = createdata()
myperceptron = Perceptron(x, y)
w, b = myperceptron.train()
python调包：
import numpy as np
from sklearn.linear_model import Perceptron
#创建数据集
def creatdata():
x = np.array([[3, -3], [4, -3], [1, 1], [1, 2]])
y = np.array([-1, -1, 1, 1])
return x, y
#感知机模型
def MyPerceptron(x, y):
clf = Perceptron(fit_intercept=True, max_iter=50, shuffle=False) #定义感知机
clf.fit(x, y) #训练感知机
w = clf.coef_ #得到权重矩阵
b = clf.intercept_ #得到截距
return w, b
if __name__ == '__main__':
x, y = creatdata()
w, b = MyPerceptron(x, y)
print('最终训练得到的w和b为：', w, ',', b)
c++实现：
#include <iostream>
#include <vector>
//创建数据集
void createdata(std::vector<std::vector<float>>& x, std::vector<float>& y)
{
x = { {3, -3}, {4, -3},{ 1, 1}, {1, 2} };
y = { -1, -1, 1, 1 };
}
//感知机模型
class Perceptron
{
public:
Perceptron(std::vector<std::vector<float>> x, std::vector<float> y)
{
m_x = x;
m_y = y;
m_w.resize(m_x[0].size(), 0);
m_b = 0;
}
float sign(std::vector<float> w, float b, std::vector<float> x)
{
float y = b;
for (size_t i = 0; i < w.size(); i++)
{
y += w[i] * x[i];
}
return y;
}
void update(float label_i, std::vector<float> data_i)
{
for (size_t i = 0; i < m_w.size(); i++)
{
m_w[i] += label_i * data_i[i];
}
m_b += label_i;
}
void train()
{
bool isFind = false;
while (!isFind)
{
float count = 0;
for (size_t i = 0; i < m_x.size(); i++)
{
float tmp_y = sign(m_w, m_b, m_x[i]);
if (tmp_y*m_y[i] <= 0) //如果误分类
{
++count;
update(m_y[i], m_x[i]);
}
}
if (count == 0)
{
std::cout << "最终训练得到的w为：";
for (auto i : m_w) std::cout << i << " ";
std::cout << "\n最终训练得到的b为：";
std::cout << m_b << "\n";
isFind = true;
}
}
}
private:
std::vector<std::vector<float>> m_x;
std::vector<float> m_y;
std::vector<float> m_w;
float m_b;
};
int main(float argc, char* argv[])
{
std::vector<std::vector<float>> x;
std::vector<float> y;
createdata(x, y);
Perceptron myperceptron = Perceptron(x, y);
myperceptron.train();
system("pause");
return EXIT_SUCCESS;
}
