一元一次函数感知器： rosenblatt感知器

1. 数据文件

先写一个py文件，定义一个数据集：

import numpy as np

def get_beans(counts):
	xs = np.random.rand(counts)
	xs = np.sort(xs)
	ys = [1.2*x+np.random.rand()/10 for x in xs]
	return xs,ys

函数说明：

# np.random.rand(counts)创建counts个服从“0~1”均匀分
# 布的随机样本值。随机样本取值范围是[0,1)，不包括1。
counts = 5
xs = np.random.rand(counts)
[0.70491788 0.42003216 0.65673763 0.76751605 0.47131048]

# np.sort(a,axis)函数的作用是对给定的数组的元素进行排序
# a：需要排序的数组
# axis：指定按什么排序，默认axis = 1 按行排序， axis = 0 按列排序
xs = np.sort(xs)
[0.42003216 0.47131048 0.65673763 0.70491788 0.76751605]

# 对于生成一个列表ys，其和xs的关系为1.2*xs+np.random.rand()/10
ys = [1.2*x+np.random.rand()/10 for x in xs]
[0.05303335450122437, 0.6920008171892412, 0.8394550531216658, 0.8587572065335468, 1.1020587170327967]

2. rosenblatt感知器

创建一个py文件，命名为rosenblatt.py

import dataset
from matplotlib import pyplot as plt
xs, ys = dataset.get_beans(100)

plt.title("size-toxicity function", fontsize = 12)
plt.xlabel("bean size")
plt.ylabel("toxicity")
plt.scatter(xs,ys)   # scatter用于绘制散点图


# 初始预测这个散点图的模型为y = 0.5*x
w = 0.5

for m in range(100):
	for i in range(100):
		x = xs[i]
		y = ys[i]
		y_pre = w * x
		e = y - y_pre
		alpha = 0.05
		w = w + alpha * e * x

# 这里 dataset里的随机数是numpy实现的
# 当一个数与一个数组进行乘法运算的时候，
# 自动把这个数广播给数组中的每一个元素，进行相乘
# 这就是numpy里的广播机制
y_pre = w * xs
print(y_pre)


plt.plot(xs, y_pre)
plt.show()

3. 预测结果

D:\lesson1>python rosenblatt.py
[0.03780218 0.04059785 0.0477705  0.0855943  0.12874181 0.13685395
 0.14489358 0.15819581 0.15873595 0.17213861 0.18243123 0.18766552
 0.19346648 0.19838957 0.20573188 0.20668577 0.24096686 0.24422348
 0.24707373 0.25030397 0.26286694 0.26648086 0.31443808 0.35684708
 0.3591003  0.39489619 0.39817362 0.40656083 0.41692886 0.41976208
 0.42143049 0.43370606 0.44166334 0.44266134 0.44797852 0.50634899
 0.54419973 0.54498475 0.56182277 0.5891967  0.59048749 0.61811849
 0.63202426 0.63321798 0.63505258 0.64797476 0.64997838 0.65315589
 0.67106602 0.67516753 0.67553516 0.7157855  0.73822292 0.75061685
 0.75788042 0.76475806 0.79805265 0.80329171 0.81531963 0.81951605
 0.83349578 0.84459298 0.84777837 0.84847466 0.84959395 0.85415659
 0.8768467  0.87854525 0.87910801 0.8931603  0.89518502 0.90568107
 0.92065886 0.93747641 0.94244445 0.96398146 0.97403517 1.01397134
 1.02901236 1.03930353 1.05150133 1.06272852 1.06790564 1.07580509
 1.07822583 1.08778958 1.11816681 1.12882552 1.13526779 1.14116164
 1.14473274 1.16009282 1.16715462 1.17221958 1.1767737  1.18421983
 1.19188772 1.19945019 1.21415735 1.24617466]