青衫烟雨客
似是故人来
前言
介绍
之前我们尝试过传统方式对手写数据集MNIST进行分类，这次使用卷积神经网络再次尝试一下。
卷积层利用卷积核对图像进行各种操作，在CNN中来说，卷积层大多是进行特征生成。但这样往往会造成内存吃不消，因此需要引入池化层，提取重要特征，缩小样本。
实验
数据准备
#prepare data
mnist = fetch_mldata('MNIST original')
data, target = mnist['data'], mnist['target']
one_hot_encoder = OneHotEncoder()
target = one_hot_encoder.fit_transform([[x] for x in target]).todense()
X_train, X_test, y_train, y_test = data[:60000], data[60000:], target[:60000], target[60000:]
使用的sklearn中的MNIST数据，由于tensorflow的缘故，将其中标签列进行独热编码。
建立计算图
-
准备输入
X = tf.placeholder(tf.float32,[None,784]) y = tf.placeholder(tf.float32,[None,10]) X_image = tf.reshape(X,[-1,28,28,1])
图片尺寸：28*28 -
第一个卷积层
conv1 = tf.layers.conv2d(X_image,filters=32, kernel_size=3, strides=1, padding="SAME",activation=tf.nn.relu, name="conv1")
图片尺寸：28*28，32个特征生成 -
第二个卷积层
conv2 = tf.layers.conv2d(conv1,filters=64, kernel_size=3, strides=2, padding="SAME",activation=tf.nn.relu, name="conv2")
图片尺寸：14*14，64个特征生成 -
池化层
pool3 = tf.nn.max_pool(conv2,ksize=[1, 2, 2, 1],strides=[1, 2, 2, 1],padding="VALID") pool3_flat = tf.reshape(pool3, shape=[-1, 64 * 7 * 7])
图片尺寸：7*7，64个特征
并将其合并为一列，为连接层准备 -
连接层与输出层
fc1 = tf.layers.dense(pool3_flat, 64, activation=tf.nn.relu, name="fc1") logits = tf.layers.dense(fc1, 10, name="output") Y_proba = tf.nn.softmax(logits, name="Y_proba") -
损失函数与训练
xentropy = tf.nn.softmax_cross_entropy_with_logits(logits=Y_proba, labels=y) loss = tf.reduce_mean(xentropy) optimizer = tf.train.AdamOptimizer() training_op = optimizer.minimize(loss) -
记录
now = datetime.utcnow().strftime('%Y%m%d%H%M%S') root_logdir = 'tf_logs' logdir = '{}/run-{}/'.format(root_logdir, now) loss_change = tf.summary.scalar('loss',loss) file_writer = tf.summary.FileWriter(logdir,tf.get_default_graph()) -
评估
y_pred = tf.arg_max(Y_proba,1) correct = tf.equal(tf.arg_max(y,1),y_pred) accuracy = tf.reduce_mean(tf.cast(correct, tf.float32)) -
开始训练
init = tf.global_variables_initializer() saver = tf.train.Saver() #prepare train def shuffle_batch(X, y, batch_size): rnd_idx = np.random.permutation(len(X)) n_batches = len(X) // batch_size for batch_idx in np.array_split(rnd_idx, n_batches): X_batch, y_batch = X[batch_idx], y[batch_idx] yield X_batch, y_batch #train n_epochs = 10 batch_size = 100 with tf.Session() as sess: init.run() for epoch in range(n_epochs): for X_batch, y_batch in shuffle_batch(X_train, y_train, batch_size): sess.run(training_op, feed_dict={X: X_batch, y: y_batch}) loss_batch = loss.eval(feed_dict={X: X_batch, y: y_batch}) acc_batch = accuracy.eval(feed_dict={X: X_batch, y: y_batch}) acc_test = accuracy.eval(feed_dict={X: X_test, y: y_test}) print(epoch, "Loss:", loss_batch, "Last batch accuracy:", acc_batch, "Test accuracy:", acc_test) save_path = saver.save(sess, "./my_mnist_model") save_path = saver.save(sess, "./my_mnist_model") file_writer.add_summary(loss_batch,epoch)
查看训练情况
可知训练陷入震荡状态
计算图如上所示
初步怀疑出现震荡是训练度不够，因此尝试再训练一个轮回，看看结果。
有所提升
损失函数也出现了下降
又训练了一轮，发现始终在震荡
死马当活马医一回，试着将batch_size设置为1看看
放弃，并没有好转，识别正确率维持在67%左右。
错误分析
绘制混淆矩阵
绘制错误率矩阵
可知8很容易和其他数字混淆
后记
sklearn实现
class CNN(BaseEstimator, ClassifierMixin):
"""docstring for CNN"""
def __init__(self):
self._session = None
def _build_graph(self):
X = tf.placeholder(tf.float32,[None,784])
y = tf.placeholder(tf.float32,[None,10])
X_image = tf.reshape(X,[-1,28,28,1])
conv1 = tf.layers.conv2d(X_image,filters=32, kernel_size=3, strides=1, padding="SAME",activation=tf.nn.relu, name="conv1")
conv2 = tf.layers.conv2d(conv1,filters=64, kernel_size=3, strides=2, padding="SAME",activation=tf.nn.relu, name="conv2")
pool3 = tf.nn.max_pool(conv2,ksize=[1, 2, 2, 1],strides=[1, 2, 2, 1],padding="VALID")
pool3_flat = tf.reshape(pool3, shape=[-1, 64 * 7 * 7])
fc1 = tf.layers.dense(pool3_flat, 64, activation=tf.nn.relu, name="fc1")
logits = tf.layers.dense(fc1, 10, name="output")
Y_proba = tf.nn.softmax(logits, name="Y_proba")
xentropy = tf.nn.softmax_cross_entropy_with_logits(logits=Y_proba, labels=y)
loss = tf.reduce_mean(xentropy)
optimizer = tf.train.AdamOptimizer()
training_op = optimizer.minimize(loss)
y_pred = tf.arg_max(Y_proba,1)
correct = tf.equal(tf.arg_max(y,1),y_pred)
accuracy = tf.reduce_mean(tf.cast(correct, tf.float32))
init = tf.global_variables_initializer()
saver = tf.train.Saver()
self._X, self._y = X, y
self._Y_proba, self._loss = Y_proba, loss
self._training_op, self._accuracy = training_op, accuracy
self._init, self._saver = init, saver
self._y_pred = y_pred
def close_session(self):
if self._session:
self._session.close()
def fit(self, X, y, n_epochs=100, batch_size=50):
self.close_session()
self._graph = tf.Graph()
with self._graph.as_default():
self._build_graph()
self._session = tf.Session(graph=self._graph)
with self._session.as_default() as sess:
self._init.run()
for epoch in range(n_epochs):
for X_batch, y_batch in shuffle_batch(X, y, batch_size):
sess.run(self._training_op, feed_dict={self._X: X_batch, self._y: y_batch})
acc_batch = self._accuracy.eval(feed_dict={self._X: X_batch, self._y: y_batch})
acc_test = self._accuracy.eval(feed_dict={self._X: X_test, self._y: y_test})
print(epoch,"Last batch accuracy:", acc_batch, "Test accuracy:", acc_test)
if acc_test > 0.98:
print("Stop!!")
break
def restore_model(self):
self.close_session()
self._graph = tf.Graph()
with self._graph.as_default():
self._build_graph()
self._session = tf.Session(graph=self._graph)
with self._session.as_default() as sess:
self._saver.restore(sess,'my_mnist_model')
def predict(self, X):
if not self._session:
raise NotFittedError("This %s instance is not fitted yet" % self.__class__.__name__)
with self._session.as_default() as sess:
return self._y_pred.eval(feed_dict={self._X: X})
匆忙，没有将使用到的参数变量化，以供之后使用
