多层感知器（Multilayer Perceptron,缩写MLP）是一种前向结构的人工神经网络，映射一组输入向量到一组输出向量。MLP可以被看作是一个有向图，由多个的节点层所组成，每一层都全连接到下一层。除了输入节点，每个节点都是一个带有非线性激活函数的神经元（或称处理单元）。一种被称为反向传播算法的监督学习方法常被用来训练MLP。MLP是感知器的推广，克服了感知器不能对线性不可分数据进行识别的弱点。

MNIST数据集使用多层感知机，输入层为784维，每一维对应一个输入节点。784维全连接到隐层。这个例子一共四层网络，输入层784个节点，二个隐藏层分别256个节点，输出层10个节点。所有层使用全连接方式连接。

import tensorflow as tf


from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)


# 参数
#学习率，迭代次数，batch大小
learning_rate = 0.001
training_epochs = 15
batch_size = 100
display_step = 1

# 网络参数
n_hidden_1 = 256 # 第一层的特征数（神经元数）
n_hidden_2 = 256 # 2nd layer number of features
n_input = 784 # MNIST 输入
n_classes = 10 # MNIST 类别数(0-9)

# tf 图的输入
x = tf.placeholder("float", [None, n_input])
y = tf.placeholder("float", [None, n_classes])


# 创建多层感知机模型
def multilayer_perceptron(x, weights, biases):
    # Hidden layer with RELU activation
    layer_1 = tf.add(tf.matmul(x, weights['h1']), biases['b1'])
    layer_1 = tf.nn.relu(layer_1)
    # Hidden layer with RELU activation
    layer_2 = tf.add(tf.matmul(layer_1, weights['h2']), biases['b2'])
    layer_2 = tf.nn.relu(layer_2)
    # Output layer with linear activation
    out_layer = tf.matmul(layer_2, weights['out']) + biases['out']
    return out_layer


# 权重、偏置参数
weights = {
    'h1': tf.Variable(tf.random_normal([n_input, n_hidden_1])),
    'h2': tf.Variable(tf.random_normal([n_hidden_1, n_hidden_2])),
    'out': tf.Variable(tf.random_normal([n_hidden_2, n_classes]))
}
biases = {
    'b1': tf.Variable(tf.random_normal([n_hidden_1])),
    'b2': tf.Variable(tf.random_normal([n_hidden_2])),
    'out': tf.Variable(tf.random_normal([n_classes]))
}

# 创建模型
pred = multilayer_perceptron(x, weights, biases)

# 定义 loss 和 optimizer
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=pred, labels=y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)

#初始化变量
init = tf.global_variables_initializer()




with tf.Session() as sess:
    sess.run(init)

    # 迭代次数
    for epoch in range(training_epochs):
        avg_cost = 0.
        total_batch = int(mnist.train.num_examples/batch_size)
        # Loop over all batches
        for i in range(total_batch):
            batch_x, batch_y = mnist.train.next_batch(batch_size)
            # Run optimization op (backprop) and cost op (to get loss value)
            _, c = sess.run([optimizer, cost], feed_dict={x: batch_x,
                                                          y: batch_y})
            # 计算平均误差
            avg_cost += c / total_batch
        # Display logs per epoch step
        if epoch % display_step == 0:
            print("Epoch:", '%04d' % (epoch+1), "cost=", \
                "{:.9f}".format(avg_cost))
    print( "Optimization Finished!")

    # Test model
    correct_prediction = tf.equal(tf.argmax(pred, 1), tf.argmax(y, 1))
    # Calculate accuracy
    accuracy = tf.reduce_mean(tf.cast(correct_prediction, "float"))
    print("Accuracy:", accuracy.eval({x: mnist.test.images, y: mnist.test.labels}))

参考：https://github.com/aymericdamien/TensorFlow-Examples/blob/master/notebooks/3_NeuralNetworks/multilayer_perceptron.ipynb