机器学习实战 支持向量机

#!/usr/bin/env python
# -*- coding: utf-8 -*-

from numpy import *

'''#######********************************
Non-Kernel VErsions below
#######********************************'''

def loadDataSet(fileName):
    dataMat = []; labelMat = []
    fr = open(fileName)
    for line in fr.readlines():
        lineArr = line.strip().split('\t')
        dataMat.append([float(lineArr[0]), float(lineArr[1])])
        labelMat.append(float(lineArr[2]))
    return dataMat,labelMat

def selectJrand(i,m):
    j=i #we want to select any J not equal to i
    while (j==i):
        j = int(random.uniform(0,m))
    return j

def clipAlpha(aj,H,L):
    # 考虑无约束条件0<=a2<=C的解
    # a2(new)= H, a2(new,unc) > H
    #       = a2(new,unc), L <= a2(new,unc) <= H
    #       = L, a2(new,unc) < L
    if aj > H: 
        aj = H
    if L > aj:
        aj = L
    return aj


class optStructK:
    def __init__(self,dataMatIn, classLabels, C, toler):  # Initialize the structure with the parameters 
        self.X = dataMatIn
        self.labelMat = classLabels
        self.C = C
        self.tol = toler
        self.m = shape(dataMatIn)[0]
        self.alphas = mat(zeros((self.m,1)))
        self.b = 0
        self.eCache = mat(zeros((self.m,2))) #误差缓存 #first column is valid flag#第一列为是否有效标志位
        
def calcEkK(oS, k):
    # 分类函数
    # F(X)= W.T * X + b
    # W.T=∑(aj*yj)  X = ∑(Xj*Xi.T)
    # F(Xi)= ∑((aj*yj)*(Xj*Xi.T)) + b
    fXk = float(multiply(oS.alphas,oS.labelMat).T*(oS.X*oS.X[k,:].T)) + oS.b
    Ek = fXk - float(oS.labelMat[k])#误差
    return Ek
        
def selectJK(i, oS, Ei): #寻找满足条件max|E1-E2|的乘子a2   #this is the second choice -heurstic, and calcs Ej
    maxK = -1; maxDeltaE = 0; Ej = 0
    oS.eCache[i] = [1,Ei]  #set valid #choose the alpha that gives the maximum delta E
    validEcacheList = nonzero(oS.eCache[:,0].A)[0]
    if (len(validEcacheList)) > 1:
        for k in validEcacheList:   #loop through valid Ecache values and find the one that maximizes delta E
            if k == i: continue #don't calc for i, waste of time
            Ek = calcEkK(oS, k)
            deltaE = abs(Ei - Ek)
            if (deltaE > maxDeltaE):
                maxK = k; maxDeltaE = deltaE; Ej = Ek
        return maxK, Ej
    else:#第一次循环 随机选择 #in this case (first time around) we don't have any valid eCache values
        j = selectJrand(i, oS.m)
        Ej = calcEkK(oS, j)
    return j, Ej

def updateEkK(oS, k):#after any alpha has changed update the new value in the cache
    Ek = calcEkK(oS, k)
    oS.eCache[k] = [1,Ek]
        
def innerLK(i, oS):
    Ei = calcEkK(oS, i)
    # 第一个约束条件，存在不满足KKT条件的ai,需要更新这些ai
    # 不满足KKT条件的情况:yi*ui<=1且ai<C, yi*ui>=1且ai>0, yi*ui=1且ai=C
    if ((oS.labelMat[i]*Ei < -oS.tol) and (oS.alphas[i] < oS.C)) or ((oS.labelMat[i]*Ei > oS.tol) and (oS.alphas[i] > 0)):
        j,Ej = selectJK(i, oS, Ei) #this has been changed from selectJrand
        alphaIold = oS.alphas[i].copy(); alphaJold = oS.alphas[j].copy();
        # 由第二个约束条件 ∑ai*yi = 0 得 a1(new)*y1+a2(new)*y2 = a1(old)*y1+a2(odl)*y2 = ζ
        if (oS.labelMat[i] != oS.labelMat[j]):               #当y1 != y2 ,则alphas[j] - alphas[i] = ζ
            L = max(0, oS.alphas[j] - oS.alphas[i])          #L = max(0,-ζ)
            H = min(oS.C, oS.C + oS.alphas[j] - oS.alphas[i])#H = min(C,C-ζ)
        else:                                                #当y1 = y2,则alphas[j] + alphas[i] = ζ
            L = max(0, oS.alphas[j] + oS.alphas[i] - oS.C)   #L = max(0,ζ-C)
            H = min(oS.C, oS.alphas[j] + oS.alphas[i])       #H = min(C,ζ)
        if L==H: print "L==H"; return 0
        # eta是alpha[j]的最优修改量
        # eta = η = 2*k(x1,x2)-k(x1,x1)-k(x2,x2)
        eta = 2.0 * oS.X[i,:]*oS.X[j,:].T - oS.X[i,:]*oS.X[i,:].T - oS.X[j,:]*oS.X[j,:].T
        if eta >= 0: print "eta>=0"; return 0
        #更新a2
        #a2(new,unc)=a2(old)-y2*(E1-E2)/η #无约束条件0<=a2<=C的解，即未经剪辑时的解
        oS.alphas[j] -= oS.labelMat[j]*(Ei - Ej)/eta
        # 考虑有约束条件0<=a2<=C的解
        # a2(new)= H, a2(new,unc) > H
        #       = a2(new,unc), L <= a2(new,unc) <= H
        #       = L, a2(new,unc) < L
        oS.alphas[j] = clipAlpha(oS.alphas[j],H,L)
        updateEkK(oS, j) #added this for the Ecache
        if (abs(oS.alphas[j] - alphaJold) < 0.00001): print "j not moving enough"; return 0
        # 更新a1
        # a1(new) = a1(old)+y1*y2*(a2(old)-a2(new))
        oS.alphas[i] += oS.labelMat[j]*oS.labelMat[i]*(alphaJold - oS.alphas[j])#update i by the same amount as j
        updateEkK(oS, i) #added this for the Ecache                    #the update is in the oppostie direction
        # 更新b
        # b1=b(old) - E1 - y1*(a1(new)-a1(old))k(x1,x1) - y2(a2(new)-a2(old))k(x1,x2)
        # b2=b(old) - E2 - y1*(a1(new)-a1(old))k(x1,x2) - y2(a2(new)-a2(old))k(x2,x2)
        b1 = oS.b - Ei- oS.labelMat[i]*(oS.alphas[i]-alphaIold)*oS.X[i,:]*oS.X[i,:].T - oS.labelMat[j]*(oS.alphas[j]-alphaJold)*oS.X[i,:]*oS.X[j,:].T
        b2 = oS.b - Ej- oS.labelMat[i]*(oS.alphas[i]-alphaIold)*oS.X[i,:]*oS.X[j,:].T - oS.labelMat[j]*(oS.alphas[j]-alphaJold)*oS.X[j,:]*oS.X[j,:].T
        # b = b1, 0<a1(new)<C
        #  = b2, 0<a2(new)<C
        #  = (b1+b2)/2,其他
        if (0 < oS.alphas[i]) and (oS.C > oS.alphas[i]): oS.b = b1
        elif (0 < oS.alphas[j]) and (oS.C > oS.alphas[j]): oS.b = b2
        else: oS.b = (b1 + b2)/2.0
        return 1
    else: return 0

def smoPK(dataMatIn, classLabels, C, toler, maxIter): #数据集 类别标签 常数C 容错律 最大循环数   #full Platt SMO
    oS = optStructK(mat(dataMatIn),mat(classLabels).transpose(),C,toler)
    iter = 0
    entireSet = True; alphaPairsChanged = 0
    while (iter < maxIter) and ((alphaPairsChanged > 0) or (entireSet)):
        alphaPairsChanged = 0
        if entireSet:   #go over all
            for i in range(oS.m):#self.m = shape(dataMatIn)[0]=100
                alphaPairsChanged += innerLK(i,oS)
                print "fullSet, iter: %d i:%d, pairs changed %d" % (iter,i,alphaPairsChanged)
            iter += 1
        else:#go over non-bound (railed) alphas
            nonBoundIs = nonzero((oS.alphas.A > 0) * (oS.alphas.A < C))[0]
            for i in nonBoundIs:
                alphaPairsChanged += innerLK(i,oS)
                print "non-bound, iter: %d i:%d, pairs changed %d" % (iter,i,alphaPairsChanged)
            iter += 1
        if entireSet: entireSet = False #toggle entire set loop
        elif (alphaPairsChanged == 0): entireSet = True  
        print "iteration number: %d" % iter
    return oS.b,oS.alphas

def calcWs(alphas,dataArr,classLabels):
    X = mat(dataArr); labelMat = mat(classLabels).transpose()
    m,n = shape(X)
    w = zeros((n,1))
    for i in range(m):
        w += multiply(alphas[i]*labelMat[i],X[i,:].T)
    return w

if __name__ == '__main__':
    dataArr, labelArr = loadDataSet('testSet.txt')
    b, alphas = smoPK(dataArr, labelArr, 0.6, 0.001, 40)#数据集 类别标签 常数C 容错律 最大循环数
    print 'b:',b
    print 'alphas[alphas>0]:',alphas[alphas>0]
    for i in range(100):
        if alphas[i]>0.0: print '支持向量:', dataArr[i], labelArr[i]
    ws = calcWs(alphas, dataArr, labelArr)
    print ws
    datMat = mat(dataArr)
    print datMat[0]*mat(ws)+b, labelArr[0]

testSet.txt:

3.542485    1.977398    -1
3.018896    2.556416    -1
7.551510    -1.580030    1
2.114999    -0.004466    -1
8.127113    1.274372    1
7.108772    -0.986906    1
8.610639    2.046708    1
2.326297    0.265213    -1
3.634009    1.730537    -1
0.341367    -0.894998    -1
3.125951    0.293251    -1