参数设计好之后，需要理解tensorflow存储数据的方式：使用占位符(参考tensorflow的英文文档）
# x y placeholder
x = tf.placeholder(tf.float32, [None, n_steps, n_inputs])
y = tf.placeholder(tf.float32, [None, n_classes])
设置初始权重和偏置：
# 对 weights biases 初始值的定义
weights = {
# shape (28, 128)
'in': tf.Variable(tf.random_normal([n_inputs, n_hidden_units])),
# shape (128, 10)
'out': tf.Variable(tf.random_normal([n_hidden_units, n_classes]))
}
biases = {
# shape (128, )
'in': tf.Variable(tf.constant(0.1,shape=[n_hidden_units, ])),
# shape (10, )
'out': tf.Variable(tf.constant(0.1,shape=[n_classes, ]))
}
随后，导入前两周设计后的数据：
def init_Vec():
data_vec = open('vectors.txt')
line_vec = data_vec.readlines()
vec_list = []
for s in line_vec:
num = s.split(" ")
#print(num)
#num.remove("\n")
num = list(map(float, num))
vec_list.append(num)
return vec_list
def init_Tag():
data_tag = open('tag.txt')
line_tag = data_tag.readlines()
tag_list = []
for s in line_tag:
num = int(s)
if num == 0:
tag_list.append([0, 0, 1])
if num == 1:
tag_list.append([0, 1, 0])
if num == 2:
tag_list.append([1, 0, 0])
return tag_list
RNN定义：
首先，为什么需要矩阵转换呢，我们可以得知n_steps*n_inputs是向量的长度，我们每次输入仅仅是1/n_steps的数据，而我们需要一整块向量来计算最终的结果，需要用上一次训练好的权重，偏执来计算，然后在这个基础上在进行拟合计算，可以根据lstm的结构看出。
def RNN(X, weights, biases):
# 原始的 X 是 3 维数据, 我们需要把它变成 2 维数据才能使用 weights 的矩阵乘法
X = tf.reshape(X, [-1, n_inputs])
# X_in = W*X + b
X_in = tf.matmul(X, weights['in']) + biases['in']
# X_in ==> (128 batches, 28 steps, 128 hidden) 换回3维
X_in = tf.reshape(X_in, [-1, n_steps, n_hidden_units])
# 使用 basic LSTM Cell.
lstm_cell = tf.contrib.rnn.BasicLSTMCell(n_hidden_units, forget_bias=1.0, state_is_tuple=True)
init_state = lstm_cell.zero_state(batch_size, dtype=tf.float32) # 初始化全零 state
outputs, final_state = tf.nn.dynamic_rnn(lstm_cell, X_in, initial_state=init_state, time_major=False)
results = tf.matmul(final_state[1], weights['out']) + biases['out']
return results
最后，定义main函数，即可开始训练：
with tf.Session() as sess:
sess.run(init)
step = 1
while step * batch_size < max_samples:
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 % display_step == 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 * batch_size) + ", Minibatch Loss = " + \
"{:.6f}".format(loss) + ", Training Accuracy = " + \
"{:.5f}".format(acc))
step += 1
print("Optimization Finished!")
test_len = 10000
test_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}))
现在，我们可以看一下训练的结果：
可以看出训练的效果还是不错，在自己的训练集上做测试出来结果还算满意。
网友评论