tensorflow实现的一个最基本cnn

原理可以参考

https://ujjwalkarn.me/2016/08/11/intuitive-explanation-convnets/

以及《神经网络与深度学习》

上代码：

import tensorflow as tf
from tqdm import tqdm_notebook
from tensorflow.examples.tutorials.mnist import input_data

"""
the cnn we are going to make:
conv?relu?pool?affine?relu?affine?softmax
"""

# ==fc== straightforward
def relu(X):
    return tf.maximum(X, tf.zeros_like(X))

# ==fc== fully connected layer calculator
def affine(X, W, b):
    n = X.get_shape()[0].value # number of samples
    X_flat = tf.reshape(X, [n, -1])
    return tf.matmul(X_flat, W) + b

def flatten(X, window_h, window_w, window_c, out_h, out_w, stride=1, padding=0):
    X_padded = tf.pad(X, [[0, 0], [padding, padding], [padding, padding], [0, 0]])
    windows = []
    for y in range(out_h):
        for x in range(out_w):
            window = tf.slice(X_padded, [0, y * stride, x * stride, 0], [-1, window_h, window_w, -1])
            windows.append(window)
    stacked = tf.stack(windows)  # shape : [out_h, out_w, n, filter_h, filter_w, c]
    return tf.reshape(stacked, [-1, window_c * window_w * window_h])

# ==fc==
def max_pool(X, pool_h, pool_w, padding, stride):
    n, h, w, c = [d.value for d in X.get_shape()]
    out_h = (h + 2 * padding - pool_h) // stride + 1
    out_w = (w + 2 * padding - pool_w) // stride + 1
    X_flat = flatten(X, pool_h, pool_w, c, out_h, out_w, stride, padding)
    pool = tf.reduce_max(tf.reshape(X_flat, [out_h, out_w, n, pool_h * pool_w, c]), axis=3)
    return tf.transpose(pool, [2, 0, 1, 3])

def convolution(X, W, b, padding, stride):
    n, h, w, c = map(lambda d: d.value, X.get_shape())
    filter_h, filter_w, filter_c, filter_n = [d.value for d in W.get_shape()]
    out_h = (h + 2 * padding - filter_h) // stride + 1
    out_w = (w + 2 * padding - filter_w) // stride + 1
    X_flat = flatten(X, filter_h, filter_w, filter_c, out_h, out_w, stride, padding)
    W_flat = tf.reshape(W, [filter_h * filter_w * filter_c, filter_n])
    z = tf.matmul(X_flat, W_flat) + b  # b: 1 X filter_n
    return tf.transpose(tf.reshape(z, [out_h, out_w, n, filter_n]), [2, 0, 1, 3])

def softmax(X):
    X_centered = X - tf.reduce_max(X) # to avoid overflow
    X_exp = tf.exp(X_centered)
    exp_sum = tf.reduce_sum(X_exp, axis=1)
    return tf.transpose(tf.transpose(X_exp) / exp_sum)

def accuracy(network, t):
    t_predict = tf.argmax(network, axis=1)
    t_actual = tf.argmax(t, axis=1)
    return tf.reduce_mean(tf.cast(tf.equal(t_predict, t_actual), tf.float32))

#==fc== load data
mnist = input_data.read_data_sets("/tmp/data/", one_hot=True, reshape=False)
#==fc== set batch size
batch_size = 100
#==fc== get the first batch data
example_X, example_ys = mnist.train.next_batch(batch_size)

# ==fc== create session
session = tf.InteractiveSession()

X = tf.placeholder(‘float‘, [batch_size, 28, 28, 1])
t = tf.placeholder(‘float‘, [batch_size, 10])

filter_h, filter_w, filter_c, filter_n = 5, 5, 1, 30
W1 = tf.Variable(tf.random_normal([filter_h, filter_w, filter_c, filter_n], stddev=0.01))
b1 = tf.Variable(tf.zeros([filter_n]))

conv_layer = convolution(X, W1, b1, padding=2, stride=1)

conv_activation_layer = relu(conv_layer)

pooling_layer = max_pool(conv_activation_layer, pool_h=2, pool_w=2, padding=0, stride=2)

batch_size, pool_output_h, pool_output_w, filter_n = [d.value for d in pooling_layer.get_shape()]

# number of nodes in the hidden layer
hidden_size = 100

W2 = tf.Variable(tf.random_normal([pool_output_h*pool_output_w*filter_n, hidden_size], stddev=0.01))
b2 = tf.Variable(tf.zeros([hidden_size]))

affine_layer1 = affine(pooling_layer, W2, b2)

init = tf.global_variables_initializer()
init.run()
affine_layer1.eval({X:example_X, t:example_ys})[0]

affine_activation_layer1 = relu(affine_layer1)

affine_activation_layer1.eval({X:example_X, t:example_ys})[0]
output_size = 10
W3 = tf.Variable(tf.random_normal([hidden_size, output_size], stddev=0.01))
b3 = tf.Variable(tf.zeros([output_size]))

affine_layer2 = affine(affine_activation_layer1, W3, b3)

init = tf.global_variables_initializer()
init.run()

affine_layer2.eval({X:example_X, t:example_ys})[0]

softmax_layer = softmax(affine_layer2)

softmax_layer.eval({X:example_X, t:example_ys})[0]

def cross_entropy_error(y, t):
    return -tf.reduce_mean(tf.log(tf.reduce_sum(y * t, axis=1)))

loss = cross_entropy_error(softmax_layer, t)

loss.eval({X:example_X, t:example_ys})

learning_rate = 0.1
trainer = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss)

# number of times to iterate over training data
training_epochs = 2

# number of batches
num_batch = int(mnist.train.num_examples/batch_size)
num_batch

for epoch in range(training_epochs):
    avg_cost = 0
    for _ in range(num_batch):
        train_X, train_ys = mnist.train.next_batch(batch_size)
        trainer.run(feed_dict={X:train_X, t:train_ys})
        avg_cost += loss.eval(feed_dict={X:train_X, t:train_ys}) / num_batch

    print("Epoch:", ‘%04d‘ % (epoch+1), "cost=", "{:.9f}".format(avg_cost), flush=True)

test_x = mnist.test.images[:batch_size]
test_t = mnist.test.labels[:batch_size]

accuracy(softmax_layer, t).eval(feed_dict={X:test_x, t:test_t})

session.close()