def main():
problem_name = 'packiging'
train_sets = [None] * 2
test_sets = [None] * 2
models = [None] * 2
# fully connected
train_sets[0] = lipnet_input.get_dataset_vironova_svm(problem_name, 'train', do_oversampling=False)
train_sets[0].oversample()
test_sets[0] = lipnet_input.get_dataset_vironova_svm(problem_name, 'test', do_oversampling=False)
from model_fully_connected import ModelFullyConneted
models[0] = ModelFullyConneted(verbose=False, compile_on_build=False)
# CNN
train_sets[1] = lipnet_input.get_dataset_images_keras(problem_name, 'train', (28, 28))
train_sets[1].oversample()
test_sets[1] = lipnet_input.get_dataset_images_keras(problem_name, 'test', (28, 28))
from model_cnn import ModelCNN
models[1] = ModelCNN(verbose=False, compile_on_build=False)
model_combined = ModelCombined(verbose=True)
model_combined.fit_combined(models, train_sets, nb_epoch=20)
# print confusion matrix of train set
cf = model_combined.evaluate_combined(train_sets, models)
print 'Train:'
print cf.as_str
# print confusion matrix of test set
cf = model_combined.evaluate_combined(test_sets, models)
print 'Test:'
print cf.as_str
def main():
problem_name = 'packiging'
train_set = lipnet_input.get_dataset_vironova_svm(problem_name=problem_name,
set_name='train',
do_oversampling=False,
batch_size=None)
test_set = lipnet_input.get_dataset_vironova_svm(problem_name=problem_name,
set_name='test',
do_oversampling=False,
batch_size=None)
cf = svm(train_set, test_set)
print cf.as_str
def main():
problem_name = 'packiging'
train_set = lipnet_input.get_dataset_vironova_svm(problem_name, 'train', do_oversampling=False)
test_set = lipnet_input.get_dataset_vironova_svm(problem_name, 'test', do_oversampling=False)
#fit(train_set, test_set, nb_epoch=40, verbose=True)
model = ModelFullyConneted(verbose=True)
train_set.oversample()
model.fit(train_set, nb_epoch=40)
# print confusion matrix of train set
cf = model.evaluate(train_set)
print 'Train:'
print cf.as_str
# print confusion matrix of test set
cf = model.evaluate(test_set)
print 'Test:'
print cf.as_str
def main():
problem_name = 'packiging'
train_sets = [None] * 2
test_sets = [None] * 2
models = [None] * 2
# fully connected
train_sets[0] = lipnet_input.get_dataset_vironova_svm(
problem_name, 'train', do_oversampling=False)
train_sets[0].oversample()
test_sets[0] = lipnet_input.get_dataset_vironova_svm(problem_name,
'test',
do_oversampling=False)
from model_fully_connected import ModelFullyConneted
models[0] = ModelFullyConneted(verbose=False, compile_on_build=False)
# CNN
train_sets[1] = lipnet_input.get_dataset_images_keras(
problem_name, 'train', (28, 28))
train_sets[1].oversample()
test_sets[1] = lipnet_input.get_dataset_images_keras(
problem_name, 'test', (28, 28))
from model_cnn import ModelCNN
models[1] = ModelCNN(verbose=False, compile_on_build=False)
model_combined = ModelCombined(verbose=True)
model_combined.fit_combined(models, train_sets, nb_epoch=20)
# print confusion matrix of train set
cf = model_combined.evaluate_combined(train_sets, models)
print 'Train:'
print cf.as_str
# print confusion matrix of test set
cf = model_combined.evaluate_combined(test_sets, models)
print 'Test:'
print cf.as_str
from keras.models import Sequential
from keras.layers import Dense, Activation, Dropout
import numpy as np
import lipnet_input
from confusion_matrix import ConfusionMatrix
problem_name = 'packiging'
train_set = lipnet_input.get_dataset_vironova_svm(problem_name,
'train',
do_oversampling=False)
x = train_set._df[train_set.feature_names].values
y = train_set._df[train_set._class_columns].values
model = Sequential()
model.add(Dense(output_dim=100, input_dim=x.shape[1]))
model.add(Activation('relu'))
model.add(Dropout(0.2))
model.add(Dense(output_dim=100, input_dim=100))
model.add(Activation('relu'))
model.add(Dropout(0.2))
model.add(Dense(output_dim=y.shape[1]))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
from keras.models import Sequential
from keras.layers import Dense, Activation, Dropout
import numpy as np
import lipnet_input
from confusion_matrix import ConfusionMatrix
problem_name = 'packiging'
train_set = lipnet_input.get_dataset_vironova_svm(problem_name, 'train', do_oversampling=False)
x = train_set._df[train_set.feature_names].values
y = train_set._df[train_set._class_columns].values
model = Sequential()
model.add(Dense(output_dim=100, input_dim=x.shape[1]))
model.add(Activation('relu'))
model.add(Dropout(0.2))
model.add(Dense(output_dim=100, input_dim=100))
model.add(Activation('relu'))
model.add(Dropout(0.2))
model.add(Dense(output_dim=y.shape[1]))
model.add(Activation('softmax'))
model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
class_weights_balanced = train_set.balanced_class_weights
class_weights = {}