#__author__ = 'Administrator'#AGN自编码器import numpy as npimport sklearn.preprocessing as preimport tensorflow as tffrom tensorflow.examples.tutorials.mnist import input_dataimport randomdef xavier_init(fan_in,fan_out,constant=1):    '''    实现Xaiver初始化器    :param fan_in: 输入节点数量    :param fan_out: 输出节点数量    :param constant: 常量    :return:fan_in行fan_out列的初始化参数    '''    low=-constant*np.sqrt(6.0/(fan_in+fan_out))    high=constant*np.sqrt(6.0/(fan_in+fan_out))    return tf.random_uniform([fan_in,fan_out],minval=low,maxval=high,dtype=tf.float32)class AdditiveGaussianNoiseAutoenencoder(object):    '''    函数功能：构建函数    Input:  n_input--输入变量数            n_hidden--隐含层节点数            transfer_function--隐含层激活函数，默认为softplus            optimizer--优化器，默认为Adam            scale--高斯噪声系数，默认为0.1    '''    def __init__(self,n_input,n_hidden,transfer_function=tf.nn.softplus,                 optimizer=tf.train.AdamOptimizer(),scale=0.1):        #初始化类内参数        self.n_input=n_input        self.n_hidden=n_hidden        self.transfer=transfer_function        self.optimizer=optimizer        self.scale=tf.placeholder(tf.float32)        self.trainning_scale=scale        networks_weights=self.__initialize_weights__()        self.weights=networks_weights        self.x=tf.placeholder(tf.float32,[None,self.n_input])        self.hidden=self.transfer(tf.add(tf.matmul(self.x+self.scale*tf.random_normal((n_input,)),self.weights['w1'])                                         ,self.weights['b1']))        self.reconstruction=tf.add(tf.matmul(self.hidden,self.weights['w2']),self.weights['b2'])        self.cost=0.5*tf.reduce_sum(tf.pow(tf.subtract(self.reconstruction,self.x),2.0))        self.optimizer=optimizer.minimize(self.cost)        init=tf.global_variables_initializer()        self.sess=tf.Session()        self.sess.run(init)    def __initialize_weights__(self):        '''        参数初始化函数        :return:所有初始化的参数        '''        all_weights=dict()        all_weights['w1']=tf.Variable(xavier_init(self.n_input,self.n_hidden))        all_weights['b1']=tf.Variable(tf.zeros([self.n_hidden],dtype=tf.float32))        all_weights['w2']=tf.Variable(tf.zeros([self.n_hidden,self.n_input],dtype=tf.float32))        all_weights['b2']=tf.Variable(tf.zeros([self.n_input],dtype=tf.float32))        return all_weights    def partial_fit(self,X):        '''        执行一步训练        :param X: 输入样本        :return:一步训练的损失函数        '''        cost, opt=self.sess.run((self.cost,self.optimizer),                                feed_dict={self.x:X,self.scale:self.trainning_scale})        return cost    def calc_total_cost(self,X):        '''        只计算cost量，在测试时会用到        :param X: 输入样本        :return:损失函数        '''        return self.sess.run(self.cost,feed_dict={self.x:X,self.scale:self.trainning_scale})    def transform(self,X):        '''        输出学到的高阶特征        :param X: 输入样本        :return:W1        '''        return self.sess.run(self.hidden,feed_dict={self.x:X,self.scale:self.trainning_scale})    def generate(self,hidden=None):        '''        和transform将整个自编码器拆分成两部分        :param hidden: 提取到的高阶特征        :return:重建后的结果        '''        if hidden is None:            hidden=np.random.normal(size=self.weights['b1'])        return self.sess.run(self.reconstruction,feed_dict={self.hidden:hidden})    def reconstruct(self,X):        '''        返回重建结果        :param X: 输入样本        :return:重建的输入样本        '''        return self.sess.run(self.reconstruction,feed_dict={self.x:X,self.scale:self.trainning_scale})    def getWeights(self):        '''        获取隐含层的权重W1        :return:        '''        return self.sess.run(self.weights['w1'])    def getBias(self):        '''        获取隐含层的贬值系数b1        :return:        '''        return self.sess.run(self.weights['b1'])def standard_scale(X_train,X_test):    '''    标准化训练、测试数据    :param X_train: 训练数据    :param X_test: 测试数据    :return:标准化后的训练、测试数据    '''    preprocessor=pre.StandardScaler().fit(X_train)    X_train=preprocessor.transform(X_train)    X_test=preprocessor.transform(X_test)    return X_train,X_testdef get_random_block_from_data(data,batch_size):    '''    获取随机block数据    :param data: 输入数据    :param batch_size: 输出block的size    :return:data中的一个随机块    '''    start_index=random.randint(0,len(data)-batch_size)    return data[start_index:(start_index+batch_size)]#Step 1:获取数据Mnist=input_data.read_data_sets('MNIST_data',one_hot=True)X_train, X_test=standard_scale(Mnist.train.images,Mnist.test.images)#Step 2:参数设置n_samples=int(Mnist.train.num_examples)#训练样本数trainning_epoches=20 #最大训练轮数为20batch_size=128  #block大小display_step=1  #每1轮显示一次损失#Step 3:创建一个AGNAE实例antoencoder=AdditiveGaussianNoiseAutoenencoder(n_input=784,n_hidden=200,transfer_function=tf.nn.softplus,                                               optimizer=tf.train.AdamOptimizer(learning_rate=0.001),                                               scale=0.1)#Step 4:训练for epoch in range(trainning_epoches):    avg_cost=0.    total_batch=int(n_samples/batch_size)    for i in range(total_batch):        batch_xs=get_random_block_from_data(X_train,batch_size)#获取训练样本        cost=antoencoder.partial_fit(batch_xs)        avg_cost+=cost/n_samples*batch_size    if epoch%display_step==0:        print("Epoch:",'%04d'%(epoch+1),"cost=","{:.9f}".format(avg_cost))#Step 5: 检测误差print("Total_test:"+str(antoencoder.calc_total_cost(X_test)))