GitHub:https://github.com/CreatCodeBuild/TensorFlow-and-DeepLearning-Tutorial/tree/master/Season1/16-19
本文内容:
- Regularization
- Dropout
- follow the Gradient与优化
- 学习率衰减
### load.py
# encoding:utf-8
# Python2 兼容
from __future__ import print_function, division
from scipy.io import loadmat as load
import matplotlib.pyplot as plt
import numpy as np
def reformat(samples, labels):
# 改变原始数据的形状
# 0 1 2 3 3 0 1 2
# (图片高,图片宽,通道数,图片数) -> (图片数,图片高,图片宽,通道数)
new = np.transpose(samples, (3, 0, 1, 2)).astype(np.float32)
# labels 变成 one-hot encoding, [2] -> [0, 0, 1, 0, 0, 0, 0, 0, 0, 0]
# digit 0 , represented as 10
# labels 变成 one-hot encoding, [10] -> [1, 0, 0, 0, 0, 0, 0, 0, 0, 0]
labels = np.array([x[0] for x in labels]) # slow code, whatever
one_hot_labels = []
for num in labels:
one_hot = [0.0] * 10
if num == 10:
one_hot[0] = 1.0
else:
one_hot[num] = 1.0
one_hot_labels.append(one_hot)
labels = np.array(one_hot_labels).astype(np.float32)
return new, labels
def normalize(samples):
'''
并且灰度化: 从三色通道 -> 单色通道 省内存 + 加快训练速度
(R + G + B) / 3
将图片从 0 ~ 255 线性映射到 -1.0 ~ +1.0
@samples: numpy array
'''
a = np.add.reduce(samples, keepdims=True, axis=3) # shape (图片数,图片高,图片宽,通道数)
a = a/3.0
return a/128.0 - 1.0
def distribution(labels, name):
# 查看一下每个label的分布,再画个统计图
# keys:
# 0
# 1
# 2
# ...
# 9
count = {}
for label in labels:
key = 0 if label[0] == 10 else label[0]
if key in count:
count[key] += 1
else:
count[key] = 1
x = []
y = []
for k, v in count.items():
# print(k, v)
x.append(k)
y.append(v)
y_pos = np.arange(len(x))
plt.bar(y_pos, y, align='center', alpha=0.5)
plt.xticks(y_pos, x)
plt.ylabel('Count')
plt.title(name + ' Label Distribution')
plt.show()
def inspect(dataset, labels, i):
# 显示图片看看
if dataset.shape[3] == 1:
shape = dataset.shape
dataset = dataset.reshape(shape[0], shape[1], shape[2])
print(labels[i])
plt.imshow(dataset[i])
plt.show()
train = load('../data/train_32x32.mat')
test = load('../data/test_32x32.mat')
# extra = load('../data/extra_32x32.mat')
# print('Train Samples Shape:', train['X'].shape)
# print('Train Labels Shape:', train['y'].shape)
# print('Train Samples Shape:', test['X'].shape)
# print('Train Labels Shape:', test['y'].shape)
# print('Train Samples Shape:', extra['X'].shape)
# print('Train Labels Shape:', extra['y'].shape)
train_samples = train['X']
train_labels = train['y']
test_samples = test['X']
test_labels = test['y']
# extra_samples = extra['X']
# extra_labels = extra['y']
n_train_samples, _train_labels = reformat(train_samples, train_labels)
n_test_samples, _test_labels = reformat(test_samples, test_labels)
_train_samples = normalize(n_train_samples)
_test_samples = normalize(n_test_samples)
num_labels = 10
image_size = 32
num_channels = 1
if __name__ == '__main__':
# 探索数据
pass
inspect(_train_samples, _train_labels, 1234)
# _train_samples = normalize(_train_samples)
# inspect(_train_samples, _train_labels, 1234)
# distribution(train_labels, 'Train Labels')
# distribution(test_labels, 'Test Labels')
dp.py
### dp.py
# 新的 refined api 不支持 Python2
import tensorflow as tf
from sklearn.metrics import confusion_matrix
import numpy as np
class Network():
def __init__(self, train_batch_size, test_batch_size, pooling_scale,
optimizeMethod='adam'):
'''
@num_hidden: 隐藏层的节点数量
@batch_size:因为我们要节省内存,所以分批处理数据。每一批的数据量。
'''
self.optimizeMethod = optimizeMethod;
self.train_batch_size = train_batch_size
self.test_batch_size = test_batch_size
# Hyper Parameters
self.conv_config = [] # list of dict
self.fc_config = [] # list of dict
self.conv_weights = []
self.conv_biases = []
self.fc_weights = []
self.fc_biases = []
self.pooling_scale = pooling_scale
self.pooling_stride = pooling_scale
# Graph Related
self.tf_train_samples = None
self.tf_train_labels = None
self.tf_test_samples = None
self.tf_test_labels = None
# 统计
self.merged = None
self.train_summaries = []
self.test_summaries = []
def add_conv(self, *, patch_size, in_depth, out_depth, activation='relu', pooling=False, name):
'''
This function does not define operations in the graph, but only store config in self.conv_layer_config
'''
self.conv_config.append({
'patch_size': patch_size,
'in_depth': in_depth,
'out_depth': out_depth,
'activation': activation,
'pooling': pooling,
'name': name
})
with tf.name_scope(name):
weights = tf.Variable(
tf.truncated_normal([patch_size, patch_size, in_depth, out_depth], stddev=0.1), name=name+'_weights')
biases = tf.Variable(tf.constant(0.1, shape=[out_depth]), name=name+'_biases')
self.conv_weights.append(weights)
self.conv_biases.append(biases)
def add_fc(self, *, in_num_nodes, out_num_nodes, activation='relu', name):
'''
add fc layer config to slef.fc_layer_config
'''
self.fc_config.append({
'in_num_nodes': in_num_nodes,
'out_num_nodes': out_num_nodes,
'activation': activation,
'name': name
})
with tf.name_scope(name):
weights = tf.Variable(tf.truncated_normal([in_num_nodes, out_num_nodes], stddev=0.1))
biases = tf.Variable(tf.constant(0.1, shape=[out_num_nodes]))
self.fc_weights.append(weights)
self.fc_biases.append(biases)
self.train_summaries.append(tf.histogram_summary(str(len(self.fc_weights))+'_weights', weights))
self.train_summaries.append(tf.histogram_summary(str(len(self.fc_biases))+'_biases', biases))
def apply_regularization(self, _lambda):
# L2 regularization for the fully connected parameters
regularization = 0.0
for weights, biases in zip(self.fc_weights, self.fc_biases):
regularization += tf.nn.l2_loss(weights) + tf.nn.l2_loss(biases)
# 1e5
return _lambda * regularization
# should make the definition as an exposed API, instead of implemented in the function
def define_inputs(self, *, train_samples_shape, train_labels_shape, test_samples_shape):
# 这里只是定义图谱中的各种变量
with tf.name_scope('inputs'):
self.tf_train_samples = tf.placeholder(tf.float32, shape=train_samples_shape, name='tf_train_samples')
self.tf_train_labels = tf.placeholder(tf.float32, shape=train_labels_shape, name='tf_train_labels')
self.tf_test_samples = tf.placeholder(tf.float32, shape=test_samples_shape, name='tf_test_samples')
def define_model(self):
'''
定义我的的计算图谱
'''
def model(data_flow, train=True):
'''
@data: original inputs
@return: logits
'''
# Define Convolutional Layers
for i, (weights, biases, config) in enumerate(zip(self.conv_weights, self.conv_biases, self.conv_config)):
with tf.name_scope(config['name'] + '_model'):
with tf.name_scope('convolution'):
# default 1,1,1,1 stride and SAME padding
data_flow = tf.nn.conv2d(data_flow, filter=weights, strides=[1, 1, 1, 1], padding='SAME')
data_flow = data_flow + biases
if not train:
self.visualize_filter_map(data_flow, how_many=config['out_depth'], display_size=32//(i//2+1), name=config['name']+'_conv')
if config['activation'] == 'relu':
data_flow = tf.nn.relu(data_flow)
if not train:
self.visualize_filter_map(data_flow, how_many=config['out_depth'], display_size=32//(i//2+1), name=config['name']+'_relu')
else:
raise Exception('Activation Func can only be Relu right now. You passed', config['activation'])
if config['pooling']:
data_flow = tf.nn.max_pool(
data_flow,
ksize=[1, self.pooling_scale, self.pooling_scale, 1],
strides=[1, self.pooling_stride, self.pooling_stride, 1],
padding='SAME')
if not train:
self.visualize_filter_map(data_flow, how_many=config['out_depth'], display_size=32//(i//2+1)//2, name=config['name']+'_pooling')
# Define Fully Connected Layers
for i, (weights, biases, config) in enumerate(zip(self.fc_weights, self.fc_biases, self.fc_config)):
if i == 0:
shape = data_flow.get_shape().as_list()
data_flow = tf.reshape(data_flow, [shape[0], shape[1] * shape[2] * shape[3]])
with tf.name_scope(config['name'] + 'model'):
### Dropout
if train and i == len(self.fc_weights) - 1:
data_flow = tf.nn.dropout(data_flow, 0.5, seed=4926)
###
data_flow = tf.matmul(data_flow, weights) + biases
if config['activation'] == 'relu':
data_flow = tf.nn.relu(data_flow)
elif config['activation'] is None:
pass
else:
raise Exception('Activation Func can only be Relu or None right now. You passed', config['activation'])
return data_flow
# Training computation.
logits = model(self.tf_train_samples)
with tf.name_scope('loss'):
self.loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits, self.tf_train_labels))
self.loss += self.apply_regularization(_lambda=5e-4)
self.train_summaries.append(tf.scalar_summary('Loss', self.loss))
# learning rate decay
global_step = tf.Variable(0)
lr = 0.001
dr = 0.99
learning_rate = tf.train.exponential_decay(
learning_rate=lr,
global_step=global_step*self.train_batch_size,
decay_steps=100,
decay_rate=dr,
staircase=True
)
# Optimizer.
with tf.name_scope('optimizer'):
if(self.optimizeMethod=='gradient'):
self.optimizer = tf.train \
.GradientDescentOptimizer(learning_rate) \
.minimize(self.loss)
elif(self.optimizeMethod=='momentum'):
self.optimizer = tf.train \
.MomentumOptimizer(learning_rate, 0.5) \
.minimize(self.loss)
elif(self.optimizeMethod=='adam'):
self.optimizer = tf.train \
.AdamOptimizer(learning_rate) \
.minimize(self.loss)
# Predictions for the training, validation, and test data.
with tf.name_scope('train'):
self.train_prediction = tf.nn.softmax(logits, name='train_prediction')
with tf.name_scope('test'):
self.test_prediction = tf.nn.softmax(model(self.tf_test_samples, train=False), name='test_prediction')
self.merged_train_summary = tf.merge_summary(self.train_summaries)
self.merged_test_summary = tf.merge_summary(self.test_summaries)
def run(self, data_iterator, train_samples, train_labels, test_samples, test_labels):
'''
用到Session
:data_iterator: a function that yields chuck of data
'''
# private function
def print_confusion_matrix(confusionMatrix):
print('Confusion Matrix:')
for i, line in enumerate(confusionMatrix):
print(line, line[i] / np.sum(line))
a = 0
for i, column in enumerate(np.transpose(confusionMatrix, (1, 0))):
a += (column[i] / np.sum(column)) * (np.sum(column) / 26000)
print(column[i] / np.sum(column), )
print('\n', np.sum(confusionMatrix), a)
self.writer = tf.train.SummaryWriter('./board', tf.get_default_graph())
with tf.Session(graph=tf.get_default_graph()) as session:
tf.initialize_all_variables().run()
### 训练
print('Start Training')
# batch 1000
for i, samples, labels in data_iterator(train_samples, train_labels, chunkSize=self.train_batch_size):
_, l, predictions, summary = session.run(
[self.optimizer, self.loss, self.train_prediction, self.merged_train_summary],
feed_dict={self.tf_train_samples: samples, self.tf_train_labels: labels}
)
self.writer.add_summary(summary, i)
# labels is True Labels
accuracy, _ = self.accuracy(predictions, labels)
if i % 50 == 0:
print('Minibatch loss at step %d: %f' % (i, l))
print('Minibatch accuracy: %.1f%%' % accuracy)
###
### 测试
accuracies = []
confusionMatrices = []
for i, samples, labels in data_iterator(test_samples, test_labels, chunkSize=self.test_batch_size):
result, summary = session.run(
[self.test_prediction, self.merged_test_summary],
feed_dict={self.tf_test_samples: samples}
)
self.writer.add_summary(summary, i)
accuracy, cm = self.accuracy(result, labels, need_confusion_matrix=True)
accuracies.append(accuracy)
confusionMatrices.append(cm)
print('Test Accuracy: %.1f%%' % accuracy)
print(' Average Accuracy:', np.average(accuracies))
print('Standard Deviation:', np.std(accuracies))
print_confusion_matrix(np.add.reduce(confusionMatrices))
###
def accuracy(self, predictions, labels, need_confusion_matrix=False):
'''
计算预测的正确率与召回率
@return: accuracy and confusionMatrix as a tuple
'''
_predictions = np.argmax(predictions, 1)
_labels = np.argmax(labels, 1)
cm = confusion_matrix(_labels, _predictions) if need_confusion_matrix else None
# == is overloaded for numpy array
accuracy = (100.0 * np.sum(_predictions == _labels) / predictions.shape[0])
return accuracy, cm
def visualize_filter_map(self, tensor, *, how_many, display_size, name):
print(tensor.get_shape)
filter_map = tensor[-1]
print(filter_map.get_shape())
filter_map = tf.transpose(filter_map, perm=[2, 0, 1])
print(filter_map.get_shape())
filter_map = tf.reshape(filter_map, (how_many, display_size, display_size, 1))
print(how_many)
self.test_summaries.append(tf.image_summary(name, tensor=filter_map, max_images=how_many))
### main.py
if __name__ == '__main__':
import load
from dp import Network
train_samples, train_labels = load._train_samples, load._train_labels
test_samples, test_labels = load._test_samples, load._test_labels
print('Training set', train_samples.shape, train_labels.shape)
print(' Test set', test_samples.shape, test_labels.shape)
image_size = load.image_size
num_labels = load.num_labels
num_channels = load.num_channels
def get_chunk(samples, labels, chunkSize):
'''
Iterator/Generator: get a batch of data
这个函数是一个迭代器/生成器,用于每一次只得到 chunkSize 这么多的数据
用于 for loop, just like range() function
'''
if len(samples) != len(labels):
raise Exception('Length of samples and labels must equal')
stepStart = 0 # initial step
i = 0
while stepStart < len(samples):
stepEnd = stepStart + chunkSize
if stepEnd < len(samples):
yield i, samples[stepStart:stepEnd], labels[stepStart:stepEnd]
i += 1
stepStart = stepEnd
net = Network(train_batch_size=64, test_batch_size=500, pooling_scale=2)
net.define_inputs(
train_samples_shape=(64, image_size, image_size, num_channels),
train_labels_shape=(64, num_labels),
test_samples_shape=(500, image_size, image_size, num_channels)
)
#
net.add_conv(patch_size=3, in_depth=num_channels, out_depth=16, activation='relu', pooling=False, name='conv1')
net.add_conv(patch_size=3, in_depth=16, out_depth=16, activation='relu', pooling=True, name='conv2')
net.add_conv(patch_size=3, in_depth=16, out_depth=16, activation='relu', pooling=False, name='conv3')
net.add_conv(patch_size=3, in_depth=16, out_depth=16, activation='relu', pooling=True, name='conv4')
# 4 = 两次 pooling, 每一次缩小为 1/2
# 16 = conv4 out_depth
net.add_fc(in_num_nodes=(image_size // 4) * (image_size // 4) * 16, out_num_nodes=16, activation='relu', name='fc1')
net.add_fc(in_num_nodes=16, out_num_nodes=10, activation=None, name='fc2')
net.define_model()
net.run(get_chunk, train_samples, train_labels, test_samples, test_labels)
else:
raise Exception('main.py: Should Not Be Imported!!! Must Run by "python main.py"')