def select_matching(texts, labels, words, threshold=1):
selected_content = []
for text, label in zip(texts, labels):
sent_list = text.split(' ')
word_count = [sent_list.count(w) for w in words]
word_count = np.array(word_count)
occur_times = (word_count >= 1).sum()
if occur_times >= threshold:
print(label, text, np.array(words)[word_count >= 1])
selected_content.append((label, text))
csv_save(selected_content, './data/traindata/Sentiment140/pre-processed/anew_part_of_nostem_160000.csv')
return
def select_matching(texts, labels, words, threshold=1):
selected_content = []
for text, label in zip(texts, labels):
sent_list = text.split(' ')
word_count = [sent_list.count(w) for w in words]
word_count = np.array(word_count)
occur_times = (word_count >= 1).sum()
if occur_times >= threshold:
print(label, text, np.array(words)[word_count >= 1])
selected_content.append((label, text))
csv_save(
selected_content,
'./data/traindata/Sentiment140/pre-processed/anew_part_of_nostem_160000.csv'
)
return
def preprocess_tweeets(tweets_list, tweets_labels, filename):
def isEnglish(s):
try:
s.encode('ascii')
except UnicodeEncodeError:
return False
else:
return True
processed_texts = []
for line, l in zip(tweets_list, tweets_labels):
if isEnglish(line):
processed_texts.append((l, preprocessor(line)))
# else: # print or not ?
# print(line)
os_name = get_os_name()
if os_name == 'windows':
file_dir = 'C:/Corpus/'
elif os_name == 'ubuntu':
file_dir = '/home/hs/Data/'
else:
return
csv_save(processed_texts, file_dir + filename)
def preprocess_tweeets(tweets_list, tweets_labels, filename):
def isEnglish(s):
try:
s.encode('ascii')
except UnicodeEncodeError:
return False
else:
return True
processed_texts = []
for line, l in zip(tweets_list, tweets_labels):
if isEnglish(line):
processed_texts.append((l, preprocessor(line)))
# else: # print or not ?
# print(line)
os_name = get_os_name()
if os_name == 'windows':
file_dir = 'C:/Corpus/'
elif os_name == 'ubuntu':
file_dir = '/home/hs/Data/'
else:
return
csv_save(processed_texts, file_dir + filename)
def evaluate_lenet5(learning_rate=0.1, n_epoches=200,
nkerns=[20, 50], batch_size=500):
# load data from dataset
logging.info('... loading data')
from load_data import load_qrcode
from sklearn.cross_validation import train_test_split
def upToInt(array):
array = np.mat(array)
m, n = np.shape(array)
newArray = np.zeros((m, n))
for i in range(m):
for j in range(n):
if array[i, j] > 0:
newArray[i, j] = 1
return newArray
features, labels = load_qrcode()
features = upToInt(features)
(trainData, trainLabel, testData, testY) = train_test_split(features, labels, test_size=0.2)
# trainData, trainLabel = util.load_total_data()
# testData = util.loadTestData()
train_set_x = theano.shared(np.asarray(trainData,
dtype=theano.config.floatX),
borrow=True)
train_set_y = theano.shared(np.asarray(trainLabel,
dtype=theano.config.floatX),
borrow=True)
test_set_x = theano.shared(np.asarray(testData,
dtype=theano.config.floatX),
borrow=True)
train_set_y = T.cast(train_set_y, 'int32')
n_train_batches = train_set_x.get_value(borrow=True).shape[0]
n_valid_batches = train_set_x.get_value(borrow=True).shape[0]
n_test_batches = test_set_x.get_value(borrow=True).shape[0]
n_train_batches /= batch_size
n_valid_batches /= batch_size
n_test_batches /= batch_size
rng = np.random.RandomState(23455)
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
# start-snippet-1
x = T.matrix('x')
y = T.ivector('y')
logging.info('... building the model')
layer0_input = x.reshape((batch_size, 1, 28, 28))
# Construct the first convolutional pooling layer:
# filtering reduces the image size to (28-5+1 , 28-5+1) = (24, 24)
# maxpooling reduces this further to (24/2, 24/2) = (12, 12)
# 4D output tensor is thus of shape (batch_size, nkerns[0], 12, 12)
layer0 = LeNetConvPoolLayer(
rng,
input=layer0_input,
image_shape=(batch_size, 1, 28, 28),
filter_shape=(nkerns[0], 1, 5, 5),
poolsize=(2, 2)
)
# Construct the second convolutional pooling layer
# filtering reduces the image size to (12-5+1, 12-5+1) = (8, 8)
# maxpooling reduces this further to (8/2, 8/2) = (4, 4)
# 4D output tensor is thus of shape (batch_size, nkerns[1], 4, 4)
layer1 = LeNetConvPoolLayer(
rng,
input=layer0.output,
image_shape=(batch_size, nkerns[0], 12, 12),
filter_shape=(nkerns[1], nkerns[0], 5, 5),
poolsize=(2, 2)
)
# the HiddenLayer being fully-connected, it operates on 2D matrices of
# shape (batch_size, num_pixels) (i.e matrix of rasterized images).
# This will generate a matrix of shape (batch_size, nkerns[1] * 4 * 4),
# or (500, 50 * 4 * 4) = (500, 800) with the default values.
layer2_input = layer1.output.flatten(2)
# construct a fully-connected sigmoidal layer
layer2 = HiddenLayer(
rng,
input=layer2_input,
n_in=nkerns[1] * 4 * 4,
n_out=500,
activation=T.tanh
)
# classify the values of the fully-connected sigmoidal layer
layer3 = LogisticRegression(input=layer2.output, n_in=500, n_out=10)
# the cost we minimize during training is the NLL of the model
cost = layer3.negative_log_likelihood(y)
# create a function to compute the mistakes that are made by the model
validate_model = theano.function(
[index],
layer3.errors(y),
givens={
x: train_set_x[index * batch_size: (index + 1) * batch_size],
y: train_set_y[index * batch_size: (index + 1) * batch_size],
}
)
# create a list of all model parameters to be fit by gradient descent
params = layer3.params + layer2.params + layer1.params + layer0.params
# create a list of gradients for all model parameters
grads = T.grad(cost, params)
# train_model is a function that updates the model parameters by
# SGD Since this model has many parameters, it would be tedious to
# manually create an update rule for each model parameter. We thus
# create the updates list by automatically looping over all
# (params[i], grads[i]) pairs.
updates = [
(param_i, param_i - learning_rate * grad_i)
for param_i, grad_i in zip(params, grads)
]
train_model = theano.function(
[index],
cost,
updates=updates,
givens={
x: train_set_x[index * batch_size: (index + 1) * batch_size],
y: train_set_y[index * batch_size: (index + 1) * batch_size]
}
)
validation_model = theano.function(
[index],
layer3.errors(y),
givens={
x: train_set_x[index * batch_size: (index + 1) * batch_size],
y: train_set_y[index * batch_size: (index + 1) * batch_size]
}
)
logging.info('... training')
patience = 10000
patience_increase = 2
improvement_threshold = 0.995
validation_frequency = min(n_train_batches, patience / 2)
best_validation_loss = np.inf
best_iter = 0
test_score = 0.
start_time = time.clock()
epoch = 0
done_looping = False
while (epoch < n_epoches) and (not done_looping):
epoch = epoch + 1
for minibatch_index in range(n_train_batches):
iter = (epoch - 1) * n_train_batches + minibatch_index
if iter % 100 == 0:
logging.info('training @ iter = %d' % (iter))
cost_ij = train_model(minibatch_index)
if (iter + 1) % validation_frequency == 0:
# compute zero-one loss on validation set
validation_losses = [validation_model(i) for i
in range(n_valid_batches)]
this_validation_loss = np.mean(validation_losses)
print('epoch %i, minibatch %i/%i, validation error %f %%' %
(epoch, minibatch_index + 1, n_train_batches,
this_validation_loss * 100.))
if (this_validation_loss * 100.) < 0.001:
done_looping = True
break
end_time = time.clock()
logging.info('The code for file ' +
os.path.split(__file__)[1] +
' ran for %.2fm' % ((end_time - start_time) / 60.))
# make a prediction and save file
# make a prediction
predict_model = theano.function(
inputs=[index],
outputs=layer3.predict(),
givens={
x: test_set_x[index * batch_size: (index + 1) * batch_size]
}
)
# save the result file
testLabel = np.array([])
for test_index in range(n_test_batches):
tempLabel = predict_model(test_index)
testLabel = np.hstack((testLabel, tempLabel))
from save_data import csv_save
csv_save(testLabel, './data/cnn_result.csv')
def evaluate_lenet5(learning_rate=0.1, n_epoches=200, nkerns=[20, 50], batch_size=500):
# load data from dataset
logging.info("... loading data")
from load_data import load_qrcode
from sklearn.cross_validation import train_test_split
def upToInt(array):
array = np.mat(array)
m, n = np.shape(array)
newArray = np.zeros((m, n))
for i in range(m):
for j in range(n):
if array[i, j] > 0:
newArray[i, j] = 1
return newArray
features, labels = load_qrcode()
features = upToInt(features)
(trainData, trainLabel, testData, testY) = train_test_split(features, labels, test_size=0.2)
# trainData, trainLabel = util.load_total_data()
# testData = util.loadTestData()
train_set_x = theano.shared(np.asarray(trainData, dtype=theano.config.floatX), borrow=True)
train_set_y = theano.shared(np.asarray(trainLabel, dtype=theano.config.floatX), borrow=True)
test_set_x = theano.shared(np.asarray(testData, dtype=theano.config.floatX), borrow=True)
train_set_y = T.cast(train_set_y, "int32")
n_train_batches = train_set_x.get_value(borrow=True).shape[0]
n_valid_batches = train_set_x.get_value(borrow=True).shape[0]
n_test_batches = test_set_x.get_value(borrow=True).shape[0]
n_train_batches /= batch_size
n_valid_batches /= batch_size
n_test_batches /= batch_size
rng = np.random.RandomState(23455)
# allocate symbolic variables for the data
index = T.lscalar() # index to a [mini]batch
# start-snippet-1
x = T.matrix("x")
y = T.ivector("y")
logging.info("... building the model")
layer0_input = x.reshape((batch_size, 1, 28, 28))
# Construct the first convolutional pooling layer:
# filtering reduces the image size to (28-5+1 , 28-5+1) = (24, 24)
# maxpooling reduces this further to (24/2, 24/2) = (12, 12)
# 4D output tensor is thus of shape (batch_size, nkerns[0], 12, 12)
layer0 = LeNetConvPoolLayer(
rng, input=layer0_input, image_shape=(batch_size, 1, 28, 28), filter_shape=(nkerns[0], 1, 5, 5), poolsize=(2, 2)
)
# Construct the second convolutional pooling layer
# filtering reduces the image size to (12-5+1, 12-5+1) = (8, 8)
# maxpooling reduces this further to (8/2, 8/2) = (4, 4)
# 4D output tensor is thus of shape (batch_size, nkerns[1], 4, 4)
layer1 = LeNetConvPoolLayer(
rng,
input=layer0.output,
image_shape=(batch_size, nkerns[0], 12, 12),
filter_shape=(nkerns[1], nkerns[0], 5, 5),
poolsize=(2, 2),
)
# the HiddenLayer being fully-connected, it operates on 2D matrices of
# shape (batch_size, num_pixels) (i.e matrix of rasterized images).
# This will generate a matrix of shape (batch_size, nkerns[1] * 4 * 4),
# or (500, 50 * 4 * 4) = (500, 800) with the default values.
layer2_input = layer1.output.flatten(2)
# construct a fully-connected sigmoidal layer
layer2 = HiddenLayer(rng, input=layer2_input, n_in=nkerns[1] * 4 * 4, n_out=500, activation=T.tanh)
# classify the values of the fully-connected sigmoidal layer
layer3 = LogisticRegression(input=layer2.output, n_in=500, n_out=10)
# the cost we minimize during training is the NLL of the model
cost = layer3.negative_log_likelihood(y)
# create a function to compute the mistakes that are made by the model
validate_model = theano.function(
[index],
layer3.errors(y),
givens={
x: train_set_x[index * batch_size : (index + 1) * batch_size],
y: train_set_y[index * batch_size : (index + 1) * batch_size],
},
)
# create a list of all model parameters to be fit by gradient descent
params = layer3.params + layer2.params + layer1.params + layer0.params
# create a list of gradients for all model parameters
grads = T.grad(cost, params)
# train_model is a function that updates the model parameters by
# SGD Since this model has many parameters, it would be tedious to
# manually create an update rule for each model parameter. We thus
# create the updates list by automatically looping over all
# (params[i], grads[i]) pairs.
updates = [(param_i, param_i - learning_rate * grad_i) for param_i, grad_i in zip(params, grads)]
train_model = theano.function(
[index],
cost,
updates=updates,
givens={
x: train_set_x[index * batch_size : (index + 1) * batch_size],
y: train_set_y[index * batch_size : (index + 1) * batch_size],
},
)
validation_model = theano.function(
[index],
layer3.errors(y),
givens={
x: train_set_x[index * batch_size : (index + 1) * batch_size],
y: train_set_y[index * batch_size : (index + 1) * batch_size],
},
)
logging.info("... training")
patience = 10000
patience_increase = 2
improvement_threshold = 0.995
validation_frequency = min(n_train_batches, patience / 2)
best_validation_loss = np.inf
best_iter = 0
test_score = 0.0
start_time = time.clock()
epoch = 0
done_looping = False
while (epoch < n_epoches) and (not done_looping):
epoch = epoch + 1
for minibatch_index in range(n_train_batches):
iter = (epoch - 1) * n_train_batches + minibatch_index
if iter % 100 == 0:
logging.info("training @ iter = %d" % (iter))
cost_ij = train_model(minibatch_index)
if (iter + 1) % validation_frequency == 0:
# compute zero-one loss on validation set
validation_losses = [validation_model(i) for i in range(n_valid_batches)]
this_validation_loss = np.mean(validation_losses)
print(
"epoch %i, minibatch %i/%i, validation error %f %%"
% (epoch, minibatch_index + 1, n_train_batches, this_validation_loss * 100.0)
)
if (this_validation_loss * 100.0) < 0.001:
done_looping = True
break
end_time = time.clock()
logging.info(
"The code for file " + os.path.split(__file__)[1] + " ran for %.2fm" % ((end_time - start_time) / 60.0)
)
# make a prediction and save file
# make a prediction
predict_model = theano.function(
inputs=[index], outputs=layer3.predict(), givens={x: test_set_x[index * batch_size : (index + 1) * batch_size]}
)
# save the result file
testLabel = np.array([])
for test_index in range(n_test_batches):
tempLabel = predict_model(test_index)
testLabel = np.hstack((testLabel, tempLabel))
from save_data import csv_save
csv_save(testLabel, "./data/cnn_result.csv")
from load_data import load_processed_data
from qrcode_generator import to_qrcode
import numpy as np
texts, labels = load_processed_data(data_type='train', stem=False)
feature_vec = []
i = 0
for text, label in zip(texts, labels):
text_qrcode = to_qrcode(text)
text_qrcode = np.array(list(text_qrcode.getdata()))
text_qrcode[text_qrcode > 0] = 1
feature_vec.append(np.append(label, text_qrcode))
from save_data import csv_save
csv_save(feature_vec, './data/traindata/qrcode_20000.csv')