tensorflow v2.0入门教程——04逻辑回归

喜欢ヅ旅行 2023-08-17 17:44 110阅读 0赞

[个人博客][Link 1]  
本教程需先理解[逻辑回归原理][Link 2]。  
采用MNIST手写数字数据集，60000张训练图片，10000张测试图片，图片大小为28\*28，像素值为0到255。

## 逻辑回归 ##

import tensorflow as tf
    import numpy as np
    from tensorflow.keras.datasets import mnist
    
    # MNIST数据集参数
    num_classes = 10  # 数字0到9, 10类
    num_features = 784  # 28*28
    
    # 训练参数
    learing_rate = 0.01
    training_steps = 1000
    batch_size = 256
    display_step =50
    
    # 预处理数据集
    (x_train, y_train), (x_test, y_test) = mnist.load_data()
    # 转为float32
    x_train, x_test = np.array(x_train, np.float32), np.array(x_test, np.float32)
    # 转为一维向量
    x_train, x_test = x_train.reshape([-1, num_features]), x_test.reshape([-1, num_features])
    # [0, 255] 到 [0, 1]
    x_train, x_test = x_train / 255, x_test / 255
    
    # tf.data.Dataset.from_tensor_slices 是使用x_train, y_train构建数据集
    train_data = tf.data.Dataset.from_tensor_slices((x_train, y_train))
    # 将数据集打乱，并设置batch_size大小
    train_data = train_data.repeat().shuffle(5000).batch(batch_size).prefetch(1)
    
    # 权重[748, 10]，图片大小28*28，类数
    W = tf.Variable(tf.ones([num_features, num_classes]), name="weight")
    # 偏置[10]，共10类
    b = tf.Variable(tf.zeros([num_classes]), name="bias")
    
    # 逻辑回归函数
    def logistic_regression(x):
        return tf.nn.softmax(tf.matmul(x, W) + b)
    
    # 损失函数
    def cross_entropy(y_pred, y_true):
        # tf.one_hot()函数的作用是将一个值化为一个概率分布的向量
        y_true = tf.one_hot(y_true, depth=num_classes)
        # tf.clip_by_value将y_pred的值控制在1e-9和1.0之间
        y_pred = tf.clip_by_value(y_pred, 1e-9, 1.0)
        return tf.reduce_mean(-tf.reduce_sum(y_true * tf.math.log(y_pred)))
    
    # 计算精度
    def accuracy(y_pred, y_true):
        # tf.cast作用是类型转换
        correct_prediction = tf.equal(tf.argmax(y_pred, 1), tf.cast(y_true, tf.int64))
        return tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
    
    # 优化器采用随机梯度下降
    optimizer = tf.optimizers.SGD(learning_rate)
    
    # 梯度下降
    def run_optimization(x, y):
        with tf.GradientTape() as g:
            pred = logistic_regression(x)
            loss = cross_entropy(pred, y)
        # 计算梯度
        gradients = g.gradient(loss, [W, b])
        # 更新梯度
        optimizer.apply_gradients(zip(gradients, [W, b]))
    
    # 开始训练
    for step, (batch_x, batch_y) in enumerate(train_data.take(training_steps), 1):
        run_optimization(batch_x, batch_y)
        if step % display_step == 0:
            pred = logistic_regression(batch_x)
            loss = cross_entropy(pred, batch_y)
            acc = accuracy(pred, batch_y)
            print("step: %i, loss: %f, accuracy: %f" % (step, loss, acc))
        
    # 测试模型的准确率
    pred = logistic_regression(x_test)
    print("Test Accuracy: %f" % accuracy(pred, y_test))

## 输出结果 ##

step: 50, loss: 1244.371582, accuracy: 0.656250
    step: 100, loss: 926.965942, accuracy: 0.792969
    step: 150, loss: 667.272644, accuracy: 0.832031
    step: 200, loss: 489.627258, accuracy: 0.871094
    step: 250, loss: 416.455811, accuracy: 0.878906
    step: 300, loss: 633.148315, accuracy: 0.796875
    step: 350, loss: 708.499329, accuracy: 0.816406
    step: 400, loss: 567.255005, accuracy: 0.765625
    step: 450, loss: 418.291779, accuracy: 0.890625
    step: 500, loss: 596.595642, accuracy: 0.824219
    step: 550, loss: 718.982849, accuracy: 0.746094
    step: 600, loss: 785.499329, accuracy: 0.781250
    step: 650, loss: 495.821411, accuracy: 0.847656
    step: 700, loss: 544.291626, accuracy: 0.871094
    step: 750, loss: 557.276123, accuracy: 0.867188
    step: 800, loss: 588.374207, accuracy: 0.843750
    step: 850, loss: 826.526855, accuracy: 0.804688
    step: 900, loss: 515.780884, accuracy: 0.851562
    step: 950, loss: 514.978210, accuracy: 0.855469
    step: 1000, loss: 580.234985, accuracy: 0.843750
    Test Accuracy: 0.826200

[个人博客][Link 1]

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dhcmRzZXB0ZW1iZXI_size_16_color_FFFFFF_t_70]

[Link 1]: http://www.wardseptember.top/
[Link 2]: https://blog.csdn.net/weixin_39445556/article/details/83930186
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3dhcmRzZXB0ZW1iZXI_size_16_color_FFFFFF_t_70]: /images/20230808/5e56aa2664c24782b694cfffbac56811.png