AdaBoost算法
当做重要决定时,大家可能都会考虑吸收多个专家而不是一个人的意见。机器学习处理问题时又何尝不是如此?这就是元算法(meta-algorithm)背后的思路。元算法是对其他算法进行组合的一种方式。接下里我们集中关注一个称作AdaBoost的元算法。
AdaBoost算法是集成学习的一种。集成学习就是构建多个“基学习器”,之后再将它们结合来完成学习任务的方法。集成学习通过将多个学习器进行综合,其性能通常比单个学习器要好。我们之前提过的随机森林就是集成学习的一种。
算法原理[2]
Adaboost算法:先从初始训练集合中训练出一个基学习器,再根据基学习器的表现对训练样本的权重进行调整,使得先前基学习器做错的样本在后续得到更多的关注,然后基于调整后的样本权重来训练下一个基学习器,直到基学习器的数目达到事先指定的数目M,最终将这M个学习器进行加权组合。
首先我们假设给定一个二分类的训练数据集:
T={(x1,y1),(x2,y2),⋯(xn,yn)} T = { ( x 1 , y 1 ) , ( x 2 , y 2 ) , ⋯ ( x n , y n ) }
其中 xϵRn x ϵ R n , yϵ{−1,+1} y ϵ { − 1 , + 1 } 。
初始化样本的权重为:
D1={w11,w12⋯w1N} D 1 = { w 11 , w 12 ⋯ w 1 N } , w1i=1N w 1 i = 1 N
第M个基分类器的样本权重为:
Dm={wm1,wm2⋯wmN} D m = { w m 1 , w m 2 ⋯ w m N } , ∑Ni=1wmi=1 ∑ i = 1 N w m i = 1
构建M个基学习器,最终的学习器即为基学习器的线性组合:
f(x)=∑Mi=1αiGi(x) f ( x ) = ∑ i = 1 M α i G i ( x )
其中 αi α i 为第 i i 个基学习器的系数, 为第 i i 个基学习器
在训练集合中的分类误差率为:
em=∑Ni=1P(Gm(xi)≠yi)=∑Ni=1wmiI(Gm(xi)≠yi) e m = ∑ i = 1 N P ( G m ( x i ) ≠ y i ) = ∑ i = 1 N w m i I ( G m ( x i ) ≠ y i )
在我们定义的基学习器中 em<0.5 e m < 0.5 。
再定义损失函数为指数损失函数
L(y,f(x))=Ex∼D[exp(−yf(x))] L ( y , f ( x ) ) = E x ∼ D [ e x p ( − y f ( x ) ) ]
其中y 是样本的实际类别, f(x) f ( x ) 是预测的类别,样本 x x 的权重服从D分布。E代表求期望。
定义了损失函数后,再来进行基分类器 和系数 αi α i 的求取
第一个分类器 G1(x) G 1 ( x ) 是直接将基学习算法用于初始数据分布求得,之后不断迭代,生成 αi α i 和 Gm G m 。当第m个基分类器产生后,我们应该使得其在数据集第m轮样本权重基础上的指数损失最小,即
L(αm,Gm(x))=argmin L ( α m , G m ( x ) ) = a r g m i n Ex∼Dm[exp(−yαmGm(x))] E x ∼ D m [ e x p ( − y α m G m ( x ) ) ]
=Ex∼Dm[∑yi=Gm(xi))e−αm+∑yi≠Gm(xi)e−αm] = E x ∼ D m [ ∑ y i = G m ( x i ) ) e − α m + ∑ y i ≠ G m ( x i ) e − α m ]
=e−αmP(yi=Gm(xi))+eαmP(yi≠Gm(xi)) = e − α m P ( y i = G m ( x i ) ) + e α m P ( y i ≠ G m ( x i ) )
=e−αm(1−em)+eαmem = e − α m ( 1 − e m ) + e α m e m
求解 Gm(x) G m ( x ) ,对任意的 α>0 α > 0 ,最优的 Gm(x) G m ( x ) 为
Gm(x)=argmin∑Ni=1wmiI(yi≠Gm(xi)) G m ( x ) = a r g m i n ∑ i = 1 N w m i I ( y i ≠ G m ( x i ) )
其中 wmi w m i 是第m 轮训练样本的权重。 Gm(x) G m ( x ) 就是在第m 轮中使得加权训练样本误差率最小的分类器。
得到 Gm(x) G m ( x ) 后,我们求 αm α m 。 αm α m 应使得损失函数最小,所以令下式对 αm α m 求导等于零:
∂L(αm,Gm(x))∂αm=0 ∂ L ( α m , G m ( x ) ) ∂ α m = 0
得到 αm α m 的表达式:
αm=12ln(1−emem) α m = 1 2 l n ( 1 − e m e m )
由于 em<0.5 e m < 0.5 ,所以 αm>0 α m > 0 ,且 αm α m 随着 em e m 的减小而增大。
这时候可以得到Adaboost的迭代公式:
fm(x)=fm−1(x)+αmGm(x) f m ( x ) = f m − 1 ( x ) + α m G m ( x )
训练样本 Dm D m 变化为
Dm=(wm,1,wm,2⋯wm,N) D m = ( w m , 1 , w m , 2 ⋯ w m , N )
Dm+1=(wm+1,1,wm+1,2⋯wm+1,N) D m + 1 = ( w m + 1 , 1 , w m + 1 , 2 ⋯ w m + 1 , N )
wm+1,i=wmiexp(−αmyiGm(xi))Zm w m + 1 , i = w m i e x p ( − α m y i G m ( x i ) ) Z m
其中 Zm=∑Ni=1wmiexp(−αmyiGm(xi))) Z m = ∑ i = 1 N w m i e x p ( − α m y i G m ( x i ) ) ) ,是个常数
算法代码[1]
'''
Created on Nov 28, 2010
Adaboost is short for Adaptive Boosting
@author: Peter
'''
from numpy import *
def loadSimpData():
datMat = matrix([[ 1. , 2.1],
[ 2. , 1.1],
[ 1.3, 1. ],
[ 1. , 1. ],
[ 2. , 1. ]])
classLabels = [1.0, 1.0, -1.0, -1.0, 1.0]
return datMat,classLabels
def loadDataSet(fileName): #general function to parse tab -delimited floats
numFeat = len(open(fileName).readline().split('\t')) #get number of fields
dataMat = []; labelMat = []
fr = open(fileName)
for line in fr.readlines():
lineArr =[]
curLine = line.strip().split('\t')
for i in range(numFeat-1):
lineArr.append(float(curLine[i]))
dataMat.append(lineArr)
labelMat.append(float(curLine[-1]))
return dataMat,labelMat
def stumpClassify(dataMatrix,dimen,threshVal,threshIneq):#just classify the data
retArray = ones((shape(dataMatrix)[0],1))
if threshIneq == 'lt':
retArray[dataMatrix[:,dimen] <= threshVal] = -1.0
else:
retArray[dataMatrix[:,dimen] > threshVal] = -1.0
return retArray
def buildStump(dataArr,classLabels,D):
dataMatrix = mat(dataArr); labelMat = mat(classLabels).T
m,n = shape(dataMatrix)
numSteps = 10.0; bestStump = {}; bestClasEst = mat(zeros((m,1)))
minError = inf #init error sum, to +infinity
for i in range(n):#loop over all dimensions
rangeMin = dataMatrix[:,i].min(); rangeMax = dataMatrix[:,i].max();
stepSize = (rangeMax-rangeMin)/numSteps
for j in range(-1,int(numSteps)+1):#loop over all range in current dimension
for inequal in ['lt', 'gt']: #go over less than and greater than
threshVal = (rangeMin + float(j) * stepSize)
predictedVals = stumpClassify(dataMatrix,i,threshVal,inequal)#call stump classify with i, j, lessThan
errArr = mat(ones((m,1)))
errArr[predictedVals == labelMat] = 0
weightedError = D.T*errArr #calc total error multiplied by D
#print "split: dim %d, thresh %.2f, thresh ineqal: %s, the weighted error is %.3f" % (i, threshVal, inequal, weightedError)
if weightedError < minError:
minError = weightedError
bestClasEst = predictedVals.copy()
bestStump['dim'] = i
bestStump['thresh'] = threshVal
bestStump['ineq'] = inequal
return bestStump,minError,bestClasEst
def adaBoostTrainDS(dataArr,classLabels,numIt=40):
weakClassArr = []
m = shape(dataArr)[0]
D = mat(ones((m,1))/m) #init D to all equal
aggClassEst = mat(zeros((m,1)))
for i in range(numIt):
bestStump,error,classEst = buildStump(dataArr,classLabels,D)#build Stump
#print "D:",D.T
alpha = float(0.5*log((1.0-error)/max(error,1e-16)))#calc alpha, throw in max(error,eps) to account for error=0
bestStump['alpha'] = alpha
weakClassArr.append(bestStump) #store Stump Params in Array
#print "classEst: ",classEst.T
expon = multiply(-1*alpha*mat(classLabels).T,classEst) #exponent for D calc, getting messy
D = multiply(D,exp(expon)) #Calc New D for next iteration
D = D/D.sum()
#calc training error of all classifiers, if this is 0 quit for loop early (use break)
aggClassEst += alpha*classEst
#print "aggClassEst: ",aggClassEst.T
aggErrors = multiply(sign(aggClassEst) != mat(classLabels).T,ones((m,1)))
errorRate = aggErrors.sum()/m
print ("total error:" ,errorRate)
if errorRate == 0.0: break
return weakClassArr,aggClassEst
def adaClassify(datToClass,classifierArr):
dataMatrix = mat(datToClass)#do stuff similar to last aggClassEst in adaBoostTrainDS
m = shape(dataMatrix)[0]
aggClassEst = mat(zeros((m,1)))
for i in range(len(classifierArr)):
classEst = stumpClassify(dataMatrix,classifierArr[i]['dim'],\
classifierArr[i]['thresh'],\
classifierArr[i]['ineq'])#call stump classify
aggClassEst += classifierArr[i]['alpha']*classEst
print (aggClassEst)
return sign(aggClassEst)
def plotROC(predStrengths, classLabels):
import matplotlib.pyplot as plt
cur = (1.0,1.0) #cursor
ySum = 0.0 #variable to calculate AUC
numPosClas = sum(array(classLabels)==1.0)
yStep = 1/float(numPosClas); xStep = 1/float(len(classLabels)-numPosClas)
sortedIndicies = predStrengths.argsort()#get sorted index, it's reverse
fig = plt.figure()
fig.clf()
ax = plt.subplot(111)
#loop through all the values, drawing a line segment at each point
for index in sortedIndicies.tolist()[0]:
if classLabels[index] == 1.0:
delX = 0; delY = yStep;
else:
delX = xStep; delY = 0;
ySum += cur[1]
#draw line from cur to (cur[0]-delX,cur[1]-delY)
ax.plot([cur[0],cur[0]-delX],[cur[1],cur[1]-delY], c='b')
cur = (cur[0]-delX,cur[1]-delY)
ax.plot([0,1],[0,1],'b--')
plt.xlabel('False positive rate'); plt.ylabel('True positive rate')
plt.title('ROC curve for AdaBoost horse colic detection system')
ax.axis([0,1,0,1])
plt.show()
print ("the Area Under the Curve is: ",ySum*xStep)
算法实例
这里给出一个简单的例子,又是鸢尾花,用它的前两个特征进行训练,来看看训练集上的正确率。
这里只抽取了后100个数据分为两种进行测试
datArr,labelArr = loadDataSet('iris.txt')
print(datArr[50:150],labelArr[50:150])
classifierArray,aggClassEst = adaBoostTrainDS(datArr[0:100],labelArr[0:100],20)
plotROC(aggClassEst.T,labelArr)
可能是在处理数据集或其他情况处理出错导致结果并不是很好。以后会进行改进
参考文献
[1]《机器学习实战》
[2]https://zhuanlan.zhihu.com/p/30676249