BernoulliRBM()如何进行预测呢？

[preeminence1]

最近在学neural_network，不知道用BernoulliRBM训练完数据后，如何进行预测？代码如下，请高手指教

countErrorNN = 0
indexNN = (1-ratio)*m

NNClassifier = BernoulliRBM().fit(returnMat[0:(1-ratio)*m+1,:],classLevelVector[0:(1-ratio)*m+1])
for i in range(int(testTotal)):
    preNumSVM = NNClassifier.predict(returnMat[indexNN,0:11])
    print 'The real num in line ',indexNN,'is ',classLevelVector[indexNN],', the predict num is ', preNumSVM
    if preNumSVM != classLevelVector[indexNN]:
        countErrorNN +=1
    indexNN +=1
   
accuracySVM = (testTotal-countErrorNN)/testTotal    
print 'The accuracy is ',accuracySVM

因为BernoulliRBM没有predict方法，所以会报错，请问该如何进行预测呢？

[rushiruyan]

http://scikit-learn.org/dev/auto_examples/neural_networks/plot_rbm_logistic_classification.html

[weixin_38356581]

在RMB后接一个分类器试试：
countErrorNN = 0
indexNN = (1-ratio)*m
rbm=BernoulliRBM(random_state=0,verbose=True)
logistic=linear_model.LogisticRegression()
NNClassifier =Pipeline(steps=[('rbm',rbm),('logistic',logistic)])
NNClassifier .fit(returnMat[0:(1-ratio)*m+1,:],classLevelVector[0:(1-ratio)*m+1])
for i in range(int(testTotal)):
preNumSVM = NNClassifier.predict(returnMat[indexNN,0:11])
print 'The real num in line ',indexNN,'is ',classLevelVector[indexNN],', the predict num is ', preNumSVM
if preNumSVM != classLevelVector[indexNN]:
countErrorNN +=1
indexNN +=1

accuracySVM = (testTotal-countErrorNN)/testTotal
print 'The accuracy is ',accuracySVM