-
Notifications
You must be signed in to change notification settings - Fork 0
/
rnn-mnist.py
67 lines (53 loc) · 2.61 KB
/
rnn-mnist.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
# tested on python=3..2, tf=1.7.0
# Use RNN to classify mnist
tf.set_random_seed(1)
mnist = input_data.read_data_sets("MNIST_data", one_hot=True)
n_inputs = 28 # input data in every step
n_steps = 28 # time steps
n_hidden_units = 128
batch_size = 128
n_classes = 10
# input x is 3D, 1st D is batch size, 2nd D is n_rows, 3rd D is n_cols. 2 and 3 are pixels of a image
x = tf.placeholder(tf.float32, [None, n_steps, n_inputs])
y = tf.placeholder(tf.float32, [None, n_classes])
w1 = tf.Variable(tf.random_normal([n_inputs, n_hidden_units]))
w2 = tf.Variable(tf.random_normal([n_hidden_units, n_classes]))
bias1 = tf.Variable(tf.constant(0.1, shape=[n_hidden_units,]))
bias2 = tf.Variable(tf.constant(0.1, shape=[n_classes,]))
weights = {"in": w1, "out": w2}
bias = {"in": bias1, "out": bias2}
def rnn(X, weight, bias):
# RNN consists of 3 parts: input layer, cell, output layer.
# transform input X to 2D
# X_in = tf.matmul(X, weight["in"]) + bias["in"]
# X_in = tf.reshape(X_in, [-1, n_steps, n_inputs])
# define cells
lstm_cell = tf.contrib.rnn.BasicLSTMCell(n_hidden_units)
# value of time_major depends on the shape of inputs into rnn cell
# output is a Tensor of shape [batch_size, max_time, cell.output_size], or [max_time, batch_size, cell.output_size]
output, final_state = tf.nn.dynamic_rnn(lstm_cell, X, time_major=False, dtype=tf.float32)
print(output.shape) # shape (batch, 28, 128)
print(output[:, -1, :].shape) # shape (batch, 128)
results = tf.matmul(output[:, -1, :], weights['out']) + bias['out'] # shape (batch, 10)
return results
prediction = rnn(x, weights, bias)
loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=prediction, labels=y))
train_op = tf.train.AdamOptimizer(0.001).minimize(loss)
correct = tf.equal(tf.argmax(prediction, 1), tf.argmax(y, 1))
acc = tf.reduce_mean(tf.cast(correct, tf.float32))
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
for i in range(1, 0):
batch_x, batch_y = mnist.train.next_batch(batch_size)
batch_x = batch_x.reshape([-1, n_steps, n_inputs])
sess.run(train_op, feed_dict={x: batch_x, y: batch_y})
if i % 10 == 0:
a, b = sess.run([loss, acc], feed_dict={x: batch_x, y: batch_y})
print(a, b)
if i % 50 == 0:
_x, _y = mnist.train.next_batch(batch_size)
_x = _x.reshape([-1, n_steps, n_inputs])
_loss, _acc = sess.run([loss, acc], feed_dict={x:_x, y:_y})
print("TEST: loss={}, acc={}".format(_loss, _acc))