感知器算法

import random
import math

import numpy as np

rate = 0.0001
X = [[1, 1, 1],
     [1, 0, 0],
     [1, 1, 0],
     [1, 0, 1]]

t = [1, 0, 0, 0]

m = len(X)

weights = [random.random(), random.random(), random.random()]

delta = [5, 5, 5]
epsilon = 1e-6


# 计算样本值
def predict(i):
    res = 0

    res = X[i][0] * weights[0] + X[i][1] * weights[1] + X[i][2] * weights[2]
    return res


# 激活函数
def sigmoid(x):
    return 1.0 / (1.0 + math.pow(math.e, -x))


count = 0
for j in range(1000):
    flag = False
    for i in range(3):
        if delta[i] < epsilon:
            flag = True
            break
    if flag:
        break
    count += 1
    print("迭代%d次" % count)
    for i in range(m):

        for j in range(3):
            delta[j] = 2 * (sigmoid(predict(i)) - t[i]) * sigmoid(predict(i)) * (1 - sigmoid(predict(i))) * (
                    X[i][0] + X[i][1] + X[i][2])

            weights[0] = weights[0] - rate * delta[0]
            weights[1] = weights[1] - rate * delta[1]
            weights[2] = weights[2] - rate * delta[2]

for i in range(3):
    print(weights[i])