tensorflow-神经网络识别手写数字

Posted 2022-11-25 wyply115

• 数据下载连接：http://yann.lecun.com/exdb/mnist/
• 下载t10k-images-idx3-ubyte.gz；t10k-labels-idx1-ubyte.gz；train-images-idx3-ubyte.gz；train-labels-idx1-ubyte.gz
• 简单神经网络识别手写数字

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# 数据下载连接：http://yann.lecun.com/exdb/mnist/
# 下载t10k-images-idx3-ubyte.gz；t10k-labels-idx1-ubyte.gz；train-images-idx3-ubyte.gz；train-labels-idx1-ubyte.gz

FLAGS = tf.app.flags.FLAGS
tf.app.flags.DEFINE_integer("is_train", 1, "指定程序是训练还是预测") # 指定1是训练模型，指定0是进行对测试集预测

def full_connected():
    '''
    简单神经网络对手写数字图片进行识别
    :return: None
    '''

    # 获取真实的数据
    mnist = input_data.read_data_sets("./data/mnist/", one_hot=True)

    # 1. 建立数据占位符 x[None, 784]  y[None, 10]
    with tf.variable_scope("data"):
        x = tf.placeholder(tf.float32, [None,784])
        y_true = tf.placeholder(tf.int32, [None, 10])

    # 2. 建立一个全连接层得神经网络  w[784,10]  b[10]
    with tf.variable_scope("fc_model"):
        # 随机初始化权重和偏置
        weight = tf.Variable(tf.random_normal([784,10], mean=0.0, stddev=1.0), name="w")
        bias = tf.Variable(tf.constant(0.0, shape=[10]))

        # 预测None的输出结果 [None, 784] * [784, 10] + [10] = [None, 10]
        y_predict = tf.matmul(x, weight) + bias

    # 3. 求出所有样本的损失，然后求平均值
    with tf.variable_scope("soft_cross"):
        # 求平均交叉熵损失
        loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_true, logits=y_predict))

    # 4. 梯度下降求出损失
    with tf.variable_scope("optimazer"):

        train_op = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

    # 5. 计算准确率
    with tf.variable_scope("acc"):

        equal_list = tf.equal(tf.arg_max(y_true,1), tf.arg_max(y_predict,1))
        accuracy = tf.reduce_mean(tf.cast(equal_list, tf.float32))
    # 收集变量
    tf.summary.scalar("losses", loss)
    tf.summary.scalar("acc", accuracy)
    # 高纬度变量收集
    tf.summary.histogram("weights", weight)
    tf.summary.histogram("biases", bias)

    # 定义一个合并变量得op
    merged = tf.summary.merge_all()

    # 创建一个saver保存模型
    saver = tf.train.Saver()

    # 6.定义一个初始化变量的op
    init_op = tf.global_variables_initializer()

    # 6. 开启会话进行训练
    with tf.Session() as sess:
        # 初始化变量
        sess.run(init_op)

        # 建立events文件，然后写入
        filewriter = tf.summary.FileWriter("./summary/", graph=sess.graph)

        if FLAGS.is_train == 1:

            # 迭代步数训练，更新参数预测
            for i in range(2000):
                # 取出真是存在得特征值和目标值
                mnist_x, mnist_y = mnist.train.next_batch(100)
                sess.run(train_op, feed_dict=x: mnist_x, y_true:mnist_y)

                # 写入每步训练得值
                summary = sess.run(merged, feed_dict=x: mnist_x, y_true:mnist_y)
                filewriter.add_summary(summary, i)

                print("训练第 %d 步，准确率为:%f " %(i, sess.run(accuracy, feed_dict=x: mnist_x, y_true:mnist_y)))
            # 保存模型
            saver.save(sess, "./data/ckpt/fc_model")
        else:
            # 加载模型
            saver.restore(sess, "./data/ckpt/fc_model")

            # 如果是0，做出预测
            for i in range(100):

                # 每次测试一张图片
                x_test, y_test = mnist.test.next_batch(1)

                print("第 %d 张图片，手写数字目标是 %d, 预测结果是：%d" % (
                    i,
                    tf.argmax(y_test, 1).eval(),
                    tf.argmax(sess.run(y_predict, feed_dict=x: x_test, y_true: y_test), 1).eval()
                ))

    return None


if __name__ == '__main__':
    full_connected()

• 卷积神经网络识别手写数字

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data


def weight_variables(shape):
    '''
    初始化权重
    :param shape:
    :return: w 初始化的权重
    '''
    w = tf.Variable(tf.random_normal(shape=shape, mean=0.0, stddev=1.0))
    return w

def bias_variables(shape):
    '''
    初始化偏置
    :param shape:
    :return: b 初始化的偏置
    '''
    b = tf.Variable(tf.constant(0.1, shape=shape))
    return b

def model():
    '''
    自定义卷积模型
    一卷积层：32个filter,5*5,strides=1,padding="SAME"; 池化：2*2, strides=2,padding="SAME"
    二卷积层：64个filter,5*5,strides=1,padding="SAME";池化：2*2, strides=2
    :return: None
    '''

    # 1. 准备数据占位符 x[None, 784]  y_true[None, 10]
    with tf.variable_scope("data"):
        x = tf.placeholder(tf.float32, [None, 784])
        y_true = tf.placeholder(tf.int32, [None, 10])

    # 2. 一卷积层, 卷积、激活、池化
    with tf.variable_scope("conv1"):
        # 随机初始化权重, 偏置
        w_conv1 = weight_variables([5,5,1,32])
        b_conv1 = bias_variables([32])

        # 对x改变形状[None,784] --> [None, 28, 28, 1]
        x_reshape = tf.reshape(x, [-1, 28,28,1])

        # 卷积+激活  [None, 28, 28, 1] --> [None, 28, 28, 32]
        x_relu1 = tf.nn.relu(tf.nn.conv2d(x_reshape, w_conv1, strides=[1,1,1,1], padding="SAME") + b_conv1)

        # 池化 2*2  [None, 28, 28, 32] --> [None, 14, 14, 32]
        x_pool1 = tf.nn.max_pool(x_relu1, ksize=[1,2,2,1], strides=[1,2,2,1], padding="SAME")

    # 3. 二卷积层
    with tf.variable_scope("conv2"):
        # 随机初始化权重, 偏置
        w_conv2 = weight_variables([5, 5, 32, 64])
        b_conv2 = bias_variables([64])

        # 卷积+激活  [None, 14, 14, 32] --> [None, 14, 14, 64]
        x_relu2 = tf.nn.relu(tf.nn.conv2d(x_pool1, w_conv2, strides=[1, 1, 1, 1], padding="SAME") + b_conv2)

        # 池化 2*2  [None, 14, 14, 64] --> [None, 7, 7, 64]
        x_pool2 = tf.nn.max_pool(x_relu2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding="SAME")

    # 4. 全连接层 [None,7,7,64] --> [None,7*7*64] * [7*7*64,10] + [10] = [None,10]
    with tf.variable_scope("fc"):
        # 随机初始化权重, 偏置
        w_fc = weight_variables([7*7*64, 10])
        b_fc = bias_variables([10])

        # 修改x_pool2形状
        x_fc_reshape = tf.reshape(x_pool2, [-1, 7*7*64])

        # 矩阵运算得出每个样本得10个结果
        y_predict = tf.matmul(x_fc_reshape, w_fc) + b_fc

    return x, y_true, y_predict

def conv_fc():
    # 1. 获取真实数据
    mnist = input_data.read_data_sets("./data/mnist/", one_hot=True)

    # 2. 定义模型，获得输出
    x, y_true, y_predict = model()

    # 3. 求出所有样本的损失，然后求平均值
    with tf.variable_scope("soft_cross"):
        # 求平均交叉熵损失
        loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(labels=y_true, logits=y_predict))

    # 4. 梯度下降求出损失
    with tf.variable_scope("optimazer"):
        train_op = tf.train.GradientDescentOptimizer(0.0001).minimize(loss)

    # 5. 计算准确率
    with tf.variable_scope("acc"):
        equal_list = tf.equal(tf.arg_max(y_true, 1), tf.arg_max(y_predict, 1))
        accuracy = tf.reduce_mean(tf.cast(equal_list, tf.float32))

    # 定义一个初始化变量的op
    init_op = tf.global_variables_initializer()

    # 开启会话
    with tf.Session() as sess:
        sess.run(init_op)

        # 循环训练
        for i in range(3000):
            # 取出真实数据中得特征值和目标值
            mnist_x, mnist_y =  mnist.train.next_batch(50)

            sess.run(train_op, feed_dict=x: mnist_x, y_true: mnist_y)

            print("训练第 %d 步，准确率为：%f " % (i, sess.run(accuracy, feed_dict=x: mnist_x, y_true: mnist_y)))


if __name__ == '__main__':
    conv_fc()

• 一到笔试题
  
  计算过程（通道对输出不影响）：

1. 经过一层卷积：长，H2 = (200 - 5 + 2*1)/2 +1 = 99.5 (这里不是整数是需要自己分析卷积过程，步长为2，0.5步就是1，因为padding=1，padding是填充的0无需观察，因此结果就是99)；长宽一样，因此不在计算宽。
2. 经过pooling，H2 = (99 - 3 + 2*0)/1 +1 = 97
3. 又经过一层卷积：H2 = (97 - 3 + 2*1)/1 +1 = 97，因此最终图片大小输出为97*97
   因此答案是：C. 97