• tensorflow实现多层感知机


    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()
    
    in_units = 784
    #隐含层节点数
    h1_units = 300
    
    """
    初始化参数
    """
    # 随机生成正太分布
    #tf.truncated_normal(shape, mean, stddev) :shape表示生成张量的维度,mean是均值,stddev是标准差。
    W1 = tf.Variable(tf.truncated_normal([in_units,h1_units],stddev = 0.1))
    b1 = tf.Variable(tf.zeros([h1_units]))
    W2 = tf.Variable(tf.zeros([h1_units,10]))
    b2 = tf.Variable(tf.zeros([10]))
    x =tf.placeholder(tf.float32,[None,in_units])
    keep_prob = tf.placeholder(tf.float32)
    
    #隐含层计算
    hidden1 = tf.nn.relu(tf.add(tf.matmul(x,W1),b1))
    #随即将一部分节点设为0
    hidden_drop = tf.nn.dropout(hidden1,keep_prob)
    #输出层计算
    y = tf.nn.softmax(tf.matmul(hidden_drop,W2)+b2)
    
    y_ = tf.placeholder(tf.float32,[None,10])
    #计算损失函数
    cross_entropy = tf.reduce_mean(-tf.reduce_sum(y_ * tf.log(y),reduction_indices=[1]))
    
    train_step = tf.train.AdagradOptimizer(0.3).minimize(cross_entropy)
    #初始化所有变量
    tf.global_variables_initializer().run()
    
    """
    训练集进行训练
    """
    
    for i in range(3000):
        batch_xs,batch_ys = mnist.train.next_batch(100)
        train_step.run({x:batch_xs,y_:batch_ys,keep_prob:0.75})
    
    # 测试集进行测试
    #tf.argmax(input, axis=None, name=None, dimension=None)
    #此函数是对矩阵按行或列计算最大值
    correct_prediction = tf.equal(tf.arg_max(y,1),tf.arg_max(y_,1))
    #cast(x, dtype, name=None)  将x的数据格式转化成dtype.
    accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
    
    print(accuracy.eval({x:mnist.test.images,y_:mnist.test.labels,keep_prob:1.0}))
  • 相关阅读:
    迭代器
    逻辑量词
    How to distinguish between strings in heap or literals?
    Is the “*apply” family really not vectorized?
    power of the test
    The Most Simple Introduction to Hypothesis Testing
    析构函数允许重载吗?
    C++中析构函数的作用
    在C#中的构造函数和解析函数
    c++构造函数与析构函数
  • 原文地址:https://www.cnblogs.com/jackzone/p/7448698.html
Copyright © 2020-2023  润新知