感知机实现—规律数据

最新推荐文章于 2020-08-30 22:49:31 发布

DMU_lzq1996

最新推荐文章于 2020-08-30 22:49:31 发布

阅读量222

点赞数

分类专栏：统计学习方法算法实现

本文链接：https://blog.csdn.net/DMU_lzq1996/article/details/82950054

版权

统计学习方法算法实现专栏收录该内容

4 篇文章 0 订阅

订阅专栏

from numpy import *
import matplotlib.pyplot as plt

def makeLinearSeparableData(weights, numLines):
    w = array(weights)
    numFeatures = len(weights)
    dataSet = zeros((numLines, numFeatures + 1))

    for i in range(numLines):
        x = random.rand(1, numFeatures) * 20 - 10
        innerProduct = sum(w * x)
        if innerProduct <= 0:
            dataSet[i] = append(x, -1)
        else:
            dataSet[i] = append(x, 1)

    return dataSet

def plotData(dataSet):
    fig = plt.figure()
    ax = fig.add_subplot(111)
    ax.set_title('Linear separable data set')
    plt.xlabel('X')
    plt.ylabel('Y')
    labels = array(dataSet[:,2])
    idx_1 = where(dataSet[:,2]==1)
    p1 = ax.scatter(dataSet[idx_1,0], dataSet[idx_1,1], marker='o', color='g', label=1, s=20)
    idx_2 = where(dataSet[:,2]==-1)
    p2 = ax.scatter(dataSet[idx_2,0], dataSet[idx_2,1], marker='x', color='r', label=2, s=20)
    plt.legend(loc = 'upper right')
    plt.show()

def train(dataSet, plot = False):
   
    numLines = dataSet.shape[0]
    numFeatures = dataSet.shape[1]
    w = zeros((1, numFeatures - 1))         # initialize weights
    separated = False

    i = 0;
    while not separated and i < numLines:
        if dataSet[i][-1] * sum(w * dataSet[i,0:-1]) <= 0:
            w = w + dataSet[i][-1] * dataSet[i,0:-1]
            separated = False
            i = 0;
        else:
            i += 1

    if plot == True:
        import matplotlib.pyplot as plt
        from matplotlib.lines import Line2D
        fig = plt.figure()
        ax = fig.add_subplot(111)
        ax.set_title('Linear separable data set')
        plt.xlabel('X')
        plt.ylabel('Y')
        labels = array(dataSet[:,2])
        idx_1 = where(dataSet[:,2]==1)
        p1 = ax.scatter(dataSet[idx_1,0], dataSet[idx_1,1], 
            marker='o', color='g', label=1, s=20)
        idx_2 = where(dataSet[:,2]==-1)
        p2 = ax.scatter(dataSet[idx_2,0], dataSet[idx_2,1], 
            marker='x', color='r', label=2, s=20)
        x = w[0][0] / abs(w[0][0]) * 10
        y = w[0][1] / abs(w[0][0]) * 10
        ann = ax.annotate(u"",xy=(x,y), 
            xytext=(0,0),size=20, arrowprops=dict(arrowstyle="-|>"))
        ys = (-12 * (-w[0][0]) / w[0][1], 12 * (-w[0][0]) / w[0][1])
        ax.add_line(Line2D((-12, 12), ys, linewidth=1, color='blue'))
        plt.legend(loc = 'upper right')
        plt.show()

    return w

data = makeLinearSeparableData([4,3],100)
plotData(data)
train(data, plot = True)