From bca131871bfd5c7eda5677e537011b39db0ff138 Mon Sep 17 00:00:00 2001 From: geekidentity Date: Sat, 1 Apr 2017 23:46:30 +0800 Subject: [PATCH] =?UTF-8?q?svmMLiA.py=20=E6=B3=A8=E9=87=8A?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- src/python/06.SVM/svmMLiA.py | 136 +++++++++++++++++++++++++---------- 1 file changed, 99 insertions(+), 37 deletions(-) diff --git a/src/python/06.SVM/svmMLiA.py b/src/python/06.SVM/svmMLiA.py index b25d5b5a..1694d6bd 100644 --- a/src/python/06.SVM/svmMLiA.py +++ b/src/python/06.SVM/svmMLiA.py @@ -75,7 +75,9 @@ def smoSimple(dataMatIn, classLabels, C, toler, maxIter): Args: dataMatIn: 数据集 classLabels: 类别标签 - C: 常数C + C: 松弛变量，允许有些数据点可以处于分隔面的错误一侧。 + 控制最大化间隔和保证大部分的函数间隔小于1.0这两个目标的权重。 + 可以通过调节该参数达到不同的结果。 toler: 容错率 maxIter: 退出前最大的循环次数 @@ -131,6 +133,16 @@ def smoSimple(dataMatIn, classLabels, C, toler, maxIter): def kernelTrans(X, A, kTup): # calc the kernel or transform data to a higher dimensional space + """ + 核转换函数 + Args: + X: + A: + kTup: 核函数的信息 + + Returns: + + """ m, n = shape(X) K = mat(zeros((m, 1))) if kTup[0] == 'lin': @@ -151,14 +163,23 @@ class optStruct: 建立的数据结构来保存所有的重要值 """ def __init__(self, dataMatIn, classLabels, C, toler, kTup): # Initialize the structure with the parameters + """ + + Args: + dataMatIn: + classLabels: + C: + toler: + kTup: 包含核函数信息的元组 + """ self.X = dataMatIn self.labelMat = classLabels self.C = C self.tol = toler - self.m = shape(dataMatIn)[0] + self.m = shape(dataMatIn)[0] # 数据的行数 self.alphas = mat(zeros((self.m, 1))) self.b = 0 - self.eCache = mat(zeros((self.m, 2))) # 第一列给出的是eCache是否有效的标志位，第二列给出的是实际的E值。 + self.eCache = mat(zeros((self.m, 2))) # 误差缓存，第一列给出的是eCache是否有效的标志位，第二列给出的是实际的E值。 self.K = mat(zeros((self.m, self.m))) for i in range(self.m): self.K[:, i] = kernelTrans(self.X, self.X[i, :], kTup) @@ -182,6 +203,7 @@ def calcEk(oS, k): def selectJ(i, oS, Ei): # this is the second choice -heurstic, and calcs Ej """ + 内循环的启发式方法。 选择第二个(内循环)alpha的alpha值 这里的目标是选择合适的第二个alpha值以保证每次优化中采用最大步长。 该函数的误差与第一个alpha值Ei和下标i有关。 @@ -204,8 +226,9 @@ def selectJ(i, oS, Ei): # this is the second choice -heurstic, and calcs Ej Ek = calcEk(oS, k) deltaE = abs(Ei - Ek) if (deltaE > maxDeltaE): - maxK = k; - maxDeltaE = deltaE; + # 选择具有最大步长的j + maxK = k + maxDeltaE = deltaE Ej = Ek return maxK, Ej else: # 如果是第一次循环，则随机选择一个alpha值 @@ -230,30 +253,42 @@ def updateEk(oS, k): # after any alpha has changed update the new value in the def innerL(i, oS): + """ + 内循环代码 + Args: + i: + oS: + + Returns: + + """ Ei = calcEk(oS, i) 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 = selectJ(i, oS, Ei) # this has been changed from selectJrand - alphaIold = oS.alphas[i].copy(); - alphaJold = oS.alphas[j].copy(); + alphaIold = oS.alphas[i].copy() + alphaJold = oS.alphas[j].copy() if (oS.labelMat[i] != oS.labelMat[j]): L = max(0, oS.alphas[j] - oS.alphas[i]) H = min(oS.C, oS.C + oS.alphas[j] - oS.alphas[i]) else: L = max(0, oS.alphas[j] + oS.alphas[i] - oS.C) H = min(oS.C, oS.alphas[j] + oS.alphas[i]) - if L == H: print("L==H") - return 0 + if L == H: + print("L==H") + return 0 eta = 2.0 * oS.K[i, j] - oS.K[i, i] - oS.K[j, j] # changed for kernel - if eta >= 0: print("eta>=0") - return 0 + if eta >= 0: + print("eta>=0") + return 0 oS.alphas[j] -= oS.labelMat[j] * (Ei - Ej) / eta oS.alphas[j] = clipAlpha(oS.alphas[j], H, L) - updateEk(oS, j) # added this for the Ecache - if (abs(oS.alphas[j] - alphaJold) < 0.00001): print("j not moving enough") - return 0 + updateEk(oS, j) # 更新误差缓存 + if (abs(oS.alphas[j] - alphaJold) < 0.00001): + print("j not moving enough") + return 0 oS.alphas[i] += oS.labelMat[j] * oS.labelMat[i] * (alphaJold - oS.alphas[j]) # update i by the same amount as j - updateEk(oS, i) # added this for the Ecache #the update is in the oppostie direction + updateEk(oS, i) # 更新误差缓存 #the update is in the oppostie direction b1 = oS.b - Ei - oS.labelMat[i] * (oS.alphas[i] - alphaIold) * oS.K[i, i] - oS.labelMat[j] * ( oS.alphas[j] - alphaJold) * oS.K[i, j] b2 = oS.b - Ej - oS.labelMat[i] * (oS.alphas[i] - alphaIold) * oS.K[i, j] - oS.labelMat[j] * ( @@ -269,19 +304,32 @@ def innerL(i, oS): return 0 -def smoP(dataMatIn, classLabels, C, toler, maxIter, kTup=('lin', 0)): # full Platt SMO +def smoP(dataMatIn, classLabels, C, toler, maxIter, kTup=('lin', 0)): + """ + 完整SMO算法外循环，与smoSimple有些类似，但这里的循环退出条件更多一些 + Args: + dataMatIn: + classLabels: + C: + toler: + maxIter: + kTup: + + Returns: + + """ oS = optStruct(mat(dataMatIn), mat(classLabels).transpose(), C, toler, kTup) iter = 0 - entireSet = True; + entireSet = True alphaPairsChanged = 0 while (iter < maxIter) and ((alphaPairsChanged > 0) or (entireSet)): alphaPairsChanged = 0 - if entireSet: # go over all + if entireSet: # 在数据集上遍历所有可能的alpha for i in range(oS.m): alphaPairsChanged += innerL(i, oS) print("fullSet, iter: %d i:%d, pairs changed %d" % (iter, i, alphaPairsChanged)) iter += 1 - else: # go over non-bound (railed) alphas + else: # 遍历所有的非边界alpha值，也就是不在边界0或C上的值。 nonBoundIs = nonzero((oS.alphas.A > 0) * (oS.alphas.A < C))[0] for i in nonBoundIs: alphaPairsChanged += innerL(i, oS) @@ -296,7 +344,17 @@ def smoP(dataMatIn, classLabels, C, toler, maxIter, kTup=('lin', 0)): # full Pl def calcWs(alphas, dataArr, classLabels): - X = mat(dataArr); + """ + 基于alpha计算w值 + Args: + alphas: + dataArr: + classLabels: + + Returns: + + """ + X = mat(dataArr) labelMat = mat(classLabels).transpose() m, n = shape(X) w = zeros((n, 1)) @@ -308,11 +366,11 @@ def calcWs(alphas, dataArr, classLabels): def testRbf(k1=1.3): dataArr, labelArr = loadDataSet('testSetRBF.txt') b, alphas = smoP(dataArr, labelArr, 200, 0.0001, 10000, ('rbf', k1)) # C=200 important - datMat = mat(dataArr); + datMat = mat(dataArr) labelMat = mat(labelArr).transpose() svInd = nonzero(alphas.A > 0)[0] sVs = datMat[svInd] # get matrix of only support vectors - labelSV = labelMat[svInd]; + labelSV = labelMat[svInd] print("there are %d Support Vectors" % shape(sVs)[0]) m, n = shape(datMat) errorCount = 0 @@ -323,7 +381,7 @@ def testRbf(k1=1.3): print("the training error rate is: %f" % (float(errorCount) / m)) dataArr, labelArr = loadDataSet('testSetRBF2.txt') errorCount = 0 - datMat = mat(dataArr); + datMat = mat(dataArr) labelMat = mat(labelArr).transpose() m, n = shape(datMat) for i in range(m): @@ -346,6 +404,7 @@ def img2vector(filename): def loadImages(dirName): from os import listdir hwLabels = [] + print(dirName) trainingFileList = listdir(dirName) # load the training set m = len(trainingFileList) trainingMat = zeros((m, 1024)) @@ -364,11 +423,11 @@ def loadImages(dirName): def testDigits(kTup=('rbf', 10)): dataArr, labelArr = loadImages('trainingDigits') b, alphas = smoP(dataArr, labelArr, 200, 0.0001, 10000, kTup) - datMat = mat(dataArr); + datMat = mat(dataArr) labelMat = mat(labelArr).transpose() svInd = nonzero(alphas.A > 0)[0] sVs = datMat[svInd] - labelSV = labelMat[svInd]; + labelSV = labelMat[svInd] print("there are %d Support Vectors" % shape(sVs)[0]) m, n = shape(datMat) errorCount = 0 @@ -379,7 +438,7 @@ def testDigits(kTup=('rbf', 10)): print("the training error rate is: %f" % (float(errorCount) / m)) dataArr, labelArr = loadImages('testDigits') errorCount = 0 - datMat = mat(dataArr); + datMat = mat(dataArr) labelMat = mat(labelArr).transpose() m, n = shape(datMat) for i in range(m): @@ -424,8 +483,8 @@ def selectJK(i, oS, Ei): # this is the second choice -heurstic, and calcs Ej Ek = calcEk(oS, k) deltaE = abs(Ei - Ek) if (deltaE > maxDeltaE): - maxK = k; - maxDeltaE = deltaE; + maxK = k + maxDeltaE = deltaE Ej = Ek return maxK, Ej else: # in this case (first time around) we don't have any valid eCache values @@ -444,24 +503,27 @@ def innerLK(i, oS): 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 = selectJ(i, oS, Ei) # this has been changed from selectJrand - alphaIold = oS.alphas[i].copy(); - alphaJold = oS.alphas[j].copy(); + alphaIold = oS.alphas[i].copy() + alphaJold = oS.alphas[j].copy() if (oS.labelMat[i] != oS.labelMat[j]): L = max(0, oS.alphas[j] - oS.alphas[i]) H = min(oS.C, oS.C + oS.alphas[j] - oS.alphas[i]) else: L = max(0, oS.alphas[j] + oS.alphas[i] - oS.C) H = min(oS.C, oS.alphas[j] + oS.alphas[i]) - if L == H: print("L==H") - return 0 + if L == H: + print("L==H") + return 0 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 + if eta >= 0: + print("eta>=0") + return 0 oS.alphas[j] -= oS.labelMat[j] * (Ei - Ej) / eta oS.alphas[j] = clipAlpha(oS.alphas[j], H, L) updateEk(oS, j) # added this for the Ecache - if (abs(oS.alphas[j] - alphaJold) < 0.00001): print("j not moving enough") - return 0 + if (abs(oS.alphas[j] - alphaJold) < 0.00001): + print("j not moving enough") + return 0 oS.alphas[i] += oS.labelMat[j] * oS.labelMat[i] * (alphaJold - oS.alphas[j]) # update i by the same amount as j updateEk(oS, i) # added this for the Ecache #the update is in the oppostie direction b1 = oS.b - Ei - oS.labelMat[i] * (oS.alphas[i] - alphaIold) * oS.X[i, :] * oS.X[i, :].T - oS.labelMat[j] * ( @@ -482,7 +544,7 @@ def innerLK(i, oS): def smoPK(dataMatIn, classLabels, C, toler, maxIter): # full Platt SMO oS = optStruct(mat(dataMatIn), mat(classLabels).transpose(), C, toler) iter = 0 - entireSet = True; + entireSet = True alphaPairsChanged = 0 while (iter < maxIter) and ((alphaPairsChanged > 0) or (entireSet)): alphaPairsChanged = 0