MNIST手写识别（三）

偏执的太偏执、 2022-05-23 01:50 245阅读 0赞

通过TensorFlow实现卷积神经网络，识别MNIST数据集，最终正确率99.2%左右。

## 通过TensorFlow实现卷积神经网络，识别MNIST数据集
    
    from tensorflow.examples.tutorials.mnist import input_data
    import tensorflow as tf
    mnist = input_data.read_data_sets("MNIST_data/",one_hot=True)
    sess = tf.InteractiveSession()
    
    
    # 初始化权重
    def weight_variable(shape):
        initial = tf.truncated_normal(shape, stddev=0.1)
        return tf.Variable(initial)
    # 初始化偏置
    def bias_variable(shape):
        initial = tf.constant(0.1, shape=shape)
        return tf.Variable(initial)
    
    
    # 定义卷积层
    # strides代表卷积模板移动的步长，paddin代表边界处理方式
    def conv2d(x, W):
        return tf.nn.conv2d(x, W, strides=[1,1,1,1], padding='SAME')
    # 定义池化层
    # 使用2x2的最大池化，即将一个2x2的像素块降为1x1，保留原始像素块灰度值最高的那一个
    def max_pool_2x2(x):
        return tf.nn.max_pool(x, ksize=[1,2,2,1], strides=[1,2,2,1], padding='SAME')
    
    
    # 定义输入的placeholder，x是特征，y_是真实的label
    x = tf.placeholder(tf.float32, [None, 784])
    y_ = tf.placeholder(tf.float32, [None, 10])
    x_image = tf.reshape(x, [-1,28,28,1]) #将输入的1D向量[1x784]转化为2D向量[28x28],-1代表样本数量不固定，1代表颜色通道数量
    
    # 定义第一个卷积层
    W_conv1 = weight_variable([5,5,1,32]) #卷积尺寸5x5，1个颜色通道，32个不同的卷积核
    b_conv1 = bias_variable([32])
    h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1) #使用ReLU激活函数进行非线性处理
    h_pool1 = max_pool_2x2(h_conv1) #池化操作
    
    #定义第二个卷积层
    W_conv2 = weight_variable([5,5,32,64])
    b_conv2 = bias_variable([64])
    h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
    h_pool2 = max_pool_2x2(h_conv2)
    
    # 变形，连接全连接层
    W_fc1 = weight_variable([7*7*64,1024])
    b_fc1 = bias_variable([1024])
    h_pool2_flat = tf.reshape(h_pool2, [-1,7*7*64]) #对第二个卷积层输出的tensor变形，转化为1D的向量
    h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
    
    # 使用Dropout层，减轻过拟合
    keep_prob = tf.placeholder(tf.float32) #通过一个placeholder传入keep_prob来控制
    h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
    
    # 将Dropout层的输出连接一个Softmax层，得到最后的概率输出
    W_fc2 = weight_variable([1024, 10])
    b_fc2 = bias_variable([10])
    y_conv = tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)
    
    # 定义损失函数cross_entropy
    cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y_conv),reduction_indices=[1]))
    train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)#优化器使用Adam，给予比较小的学习速率1e-4
    
    # 定义评价准确率的操作
    correct_prediction = tf.equal(tf.argmax(y_conv,1), tf.argmax(y_,1))
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
    
    
    # 开始训练
    tf.global_variables_initializer().run()
    for i in range(20000):
        batch = mnist.train.next_batch(50)
        if i%100 == 0:
            train_accuracy = accuracy.eval(feed_dict={x:batch[0], y_:batch[1], keep_prob: 1.0})
            print("step %d, training accuracy %g"%(i, train_accuracy))
        train_step.run(feed_dict={x: batch[0], y_:batch[1], keep_prob:0.5})
    
    
    # 测试数据，输出整体的分类准确率
    print("test accuracy %g"%accuracy.eval(feed_dict={x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0}))