前言

本文输入数据是MNIST，全称是Modified National Institute of Standards and Technology，是一组由这个机构搜集的手写数字扫描文件和每个文件对应标签的数据集，经过一定的修改使其适合机器学习算法读取。这个数据集可以从牛的不行的Yann LeCun教授的获取。

本系列的其他文章已经根据TensorFlow的官方教程基于MNIST数据集采用了softmax regression和CNN进行建模。为了完整性，本文对MNIST数据应用RNN模型求解，具体使用的RNN为LSTM。

关于RNN/LSTM的理论知识，可以参考

代码

# coding: utf-8# @author: 陈水平# @date：2017-02-14# # In[1]:import tensorflow as tfimport numpy as np# In[2]:sess = tf.InteractiveSession()# In[3]:from tensorflow.examples.tutorials.mnist import input_datamnist = input_data.read_data_sets('mnist/', one_hot=True)# In[4]:learning_rate = 0.001batch_size = 128n_input = 28n_steps = 28n_hidden = 128n_classes = 10x = tf.placeholder(tf.float32, [None, n_steps, n_input])y = tf.placeholder(tf.float32, [None, n_classes])# In[5]:def RNN(x, weight, biases):    # x shape: (batch_size, n_steps, n_input)    # desired shape: list of n_steps with element shape (batch_size, n_input)    x = tf.transpose(x, [1, 0, 2])    x = tf.reshape(x, [-1, n_input])    x = tf.split(0, n_steps, x)    outputs = list()    lstm = tf.nn.rnn_cell.BasicLSTMCell(n_hidden, forget_bias=1.0)    state = (tf.zeros([n_steps, n_hidden]),)*2    sess.run(state)    with tf.variable_scope("myrnn2") as scope:        for i in range(n_steps-1):            if i > 0:                scope.reuse_variables()            output, state = lstm(x[i], state)            outputs.append(output)    final = tf.matmul(outputs[-1], weight) + biases    return final# In[6]:def RNN(x, n_steps, n_input, n_hidden, n_classes):    # Parameters:    # Input gate: input, previous output, and bias    ix = tf.Variable(tf.truncated_normal([n_input, n_hidden], -0.1, 0.1))    im = tf.Variable(tf.truncated_normal([n_hidden, n_hidden], -0.1, 0.1))    ib = tf.Variable(tf.zeros([1, n_hidden]))    # Forget gate: input, previous output, and bias    fx = tf.Variable(tf.truncated_normal([n_input, n_hidden], -0.1, 0.1))    fm = tf.Variable(tf.truncated_normal([n_hidden, n_hidden], -0.1, 0.1))    fb = tf.Variable(tf.zeros([1, n_hidden]))    # Memory cell: input, state, and bias    cx = tf.Variable(tf.truncated_normal([n_input, n_hidden], -0.1, 0.1))    cm = tf.Variable(tf.truncated_normal([n_hidden, n_hidden], -0.1, 0.1))    cb = tf.Variable(tf.zeros([1, n_hidden]))    # Output gate: input, previous output, and bias    ox = tf.Variable(tf.truncated_normal([n_input, n_hidden], -0.1, 0.1))    om = tf.Variable(tf.truncated_normal([n_hidden, n_hidden], -0.1, 0.1))    ob = tf.Variable(tf.zeros([1, n_hidden]))    # Classifier weights and biases    w = tf.Variable(tf.truncated_normal([n_hidden, n_classes]))    b = tf.Variable(tf.zeros([n_classes]))    # Definition of the cell computation    def lstm_cell(i, o, state):        input_gate = tf.sigmoid(tf.matmul(i, ix) + tf.matmul(o, im) + ib)        forget_gate = tf.sigmoid(tf.matmul(i, fx) + tf.matmul(o, fm) + fb)        update = tf.tanh(tf.matmul(i, cx) + tf.matmul(o, cm) + cb)        state = forget_gate * state + input_gate * update        output_gate = tf.sigmoid(tf.matmul(i, ox) +  tf.matmul(o, om) + ob)        return output_gate * tf.tanh(state), state        # Unrolled LSTM loop    outputs = list()    state = tf.Variable(tf.zeros([batch_size, n_hidden]))    output = tf.Variable(tf.zeros([batch_size, n_hidden]))        # x shape: (batch_size, n_steps, n_input)    # desired shape: list of n_steps with element shape (batch_size, n_input)    x = tf.transpose(x, [1, 0, 2])    x = tf.reshape(x, [-1, n_input])    x = tf.split(0, n_steps, x)    for i in x:        output, state = lstm_cell(i, output, state)        outputs.append(output)    logits =tf.matmul(outputs[-1], w) + b    return logits# In[7]:pred = RNN(x, n_steps, n_input, n_hidden, n_classes)cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(pred, y))optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)correct_pred = tf.equal(tf.argmax(pred,1), tf.argmax(y,1))accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))# Initializing the variablesinit = tf.global_variables_initializer()# In[8]:# Launch the graphsess.run(init)for step in range(20000):    batch_x, batch_y = mnist.train.next_batch(batch_size)    batch_x = batch_x.reshape((batch_size, n_steps, n_input))    sess.run(optimizer, feed_dict={x: batch_x, y: batch_y})    if step % 50 == 0:        acc = sess.run(accuracy, feed_dict={x: batch_x, y: batch_y})        loss = sess.run(cost, feed_dict={x: batch_x, y: batch_y})        print "Iter " + str(step) + ", Minibatch Loss= " +               "{:.6f}".format(loss) + ", Training Accuracy= " +               "{:.5f}".format(acc)print "Optimization Finished!"# In[9]:# Calculate accuracy for 128 mnist test imagestest_len = batch_sizetest_data = mnist.test.images[:test_len].reshape((-1, n_steps, n_input))test_label = mnist.test.labels[:test_len]print "Testing Accuracy:", sess.run(accuracy, feed_dict={x: test_data, y: test_label})

输出如下：

Iter 0, Minibatch Loss= 2.540429, Training Accuracy= 0.07812Iter 50, Minibatch Loss= 2.423611, Training Accuracy= 0.06250Iter 100, Minibatch Loss= 2.318830, Training Accuracy= 0.13281Iter 150, Minibatch Loss= 2.276640, Training Accuracy= 0.13281Iter 200, Minibatch Loss= 2.276727, Training Accuracy= 0.12500Iter 250, Minibatch Loss= 2.267064, Training Accuracy= 0.16406Iter 300, Minibatch Loss= 2.234139, Training Accuracy= 0.19531Iter 350, Minibatch Loss= 2.295060, Training Accuracy= 0.12500Iter 400, Minibatch Loss= 2.261856, Training Accuracy= 0.16406Iter 450, Minibatch Loss= 2.220284, Training Accuracy= 0.17969Iter 500, Minibatch Loss= 2.276015, Training Accuracy= 0.13281Iter 550, Minibatch Loss= 2.220499, Training Accuracy= 0.14062Iter 600, Minibatch Loss= 2.219574, Training Accuracy= 0.11719Iter 650, Minibatch Loss= 2.189177, Training Accuracy= 0.25781Iter 700, Minibatch Loss= 2.195167, Training Accuracy= 0.19531Iter 750, Minibatch Loss= 2.226459, Training Accuracy= 0.18750Iter 800, Minibatch Loss= 2.148620, Training Accuracy= 0.23438Iter 850, Minibatch Loss= 2.122925, Training Accuracy= 0.21875Iter 900, Minibatch Loss= 2.065122, Training Accuracy= 0.24219...Iter 19350, Minibatch Loss= 0.001304, Training Accuracy= 1.00000Iter 19400, Minibatch Loss= 0.000144, Training Accuracy= 1.00000Iter 19450, Minibatch Loss= 0.000907, Training Accuracy= 1.00000Iter 19500, Minibatch Loss= 0.002555, Training Accuracy= 1.00000Iter 19550, Minibatch Loss= 0.002018, Training Accuracy= 1.00000Iter 19600, Minibatch Loss= 0.000853, Training Accuracy= 1.00000Iter 19650, Minibatch Loss= 0.001035, Training Accuracy= 1.00000Iter 19700, Minibatch Loss= 0.007034, Training Accuracy= 0.99219Iter 19750, Minibatch Loss= 0.000608, Training Accuracy= 1.00000Iter 19800, Minibatch Loss= 0.002913, Training Accuracy= 1.00000Iter 19850, Minibatch Loss= 0.003484, Training Accuracy= 1.00000Iter 19900, Minibatch Loss= 0.005693, Training Accuracy= 1.00000Iter 19950, Minibatch Loss= 0.001904, Training Accuracy= 1.00000Optimization Finished!Testing Accuracy: 0.992188