def main():
t = Timer()
t.reset_cpu_time()
logging.info("PREPARE DATASET")
train_X, train_Y, test_X, test_Y = load_light_dataset(
images_root_dir,
target=target,
training_set_part=0.8,
extension='jpeg')
logging.info("CREATE MODEL")
model = create_model(network['layers_and_filters'],
network['kernel_size'],
network['activation'], (IMG_SIZE, IMG_SIZE, levels),
network['dropout_rate'],
network['optimizer'],
network['learning_rate'],
output_size=train_Y.shape[1])
t.get_cpu_time("PREPARATION")
logging.info("TRAIN")
try:
if options.train == 'true':
raise Exception('Force train model')
model = models.load_model(model_name)
except:
fit_one_at_time(model, train_X, train_Y, epochs=network['epochs'])
model.save(model_name)
t.get_cpu_time("TRAIN")
logging.info("TEST on TRAIN")
score_one_at_time(model, train_X, train_Y)
t.get_cpu_time("TEST on TRAIN")
logging.info("TEST")
score_one_at_time(model, test_X, test_Y)
t.get_cpu_time("TEST")
def main():
t = Timer()
logging.info("PREPARE DATASET")
# Create train and test set
train_X, train_Y, test_X, test_Y = load_dataset_properties(
dataset, target=0, training_set_part=0.8)
train_names = train_X[:, 0]
train_X = train_X[:, 1:]
test_names = test_X[:, 0]
test_X = test_X[:, 1:]
train_size = train_X.shape[0]
test_size = test_X.shape[0]
input_size = train_X.shape[1]
output_size = train_Y.shape[1]
try:
if options.train == 'true':
raise Exception('Force train model')
model = models.load_model(model_name)
except:
# create the model
logging.info("CREATE MODEL")
model = create_model(input_size=input_size,
output_size=output_size,
n_layers=network['n_layers'],
n_neurons=network['n_neurons'],
activation_function=network['activation'],
learning_rate=network['learning_rate'],
dropout_rate=network['dropout_rate'],
optimizer=network['optimizer'])
t.reset_cpu_time()
# train the model
results = model.fit(
x=train_X,
y=train_Y,
epochs=network['epochs'],
)
t.get_cpu_time("TRAIN")
model.save(model_name)
logging.info("TEST on TRAIN")
t.reset_cpu_time()
predict_and_score(model, np.asarray(train_X).astype(np.float32), train_Y)
t.get_cpu_time("TEST on TRAIN")
logging.info("TEST")
predict_and_score(model, np.asarray(test_X).astype(np.float32), test_Y)
t.get_cpu_time("TEST")
def main():
t = Timer()
t.reset_cpu_time()
# Create traing and test set
train_X, train_Y, test_X, test_Y = load_dataset(dataset, target=target)
#dataset beging with the file name
train_X = train_X[:, 1:]
test_X = test_X[:, 1:]
train_size = train_X.shape[0]
test_size = test_X.shape[0]
input_size = train_X.shape[1]
output_size = train_Y.shape[1]
try:
if options.train == 'true':
raise Exception('Force train model')
model = models.load_model(model_name)
except:
# create the model
model = create_model(input_size=input_size,
output_size=output_size,
n_layers=network['n_layers'],
n_neurons=network['n_neurons'],
activation_function=network['activation'],
learning_rate=network['learning_rate'],
dropout_rate=network['dropout_rate'],
optimizer=network['optimizer'])
t.get_cpu_time("PREPARATION")
# train the model
results = model.fit(
x=train_X,
y=train_Y,
epochs=network['epochs'],
#validation_data= (test_X, test_Y)
)
model.save(model_name)
logging.info("TEST on TRAIN")
t.reset_cpu_time()
test(model, np.asarray(train_X).astype(np.float32), train_Y)
t.get_cpu_time("TEST on TRAIN")
logging.info("TEST")
test(model, np.asarray(test_X).astype(np.float32), test_Y)
t.get_cpu_time("TEST")
def main():
t = Timer()
t.reset_cpu_time()
logging.info("PREPARE DATASET")
train_X1, train_Y1, test_X1, test_Y1 = load_light_dataset(
images_root_dir, training_set_part=0.8, extension='jpeg')
train_X2, _, test_X2, _ = load_dataset(options.dataset_bow)
train_X3, _, test_X3, _ = load_dataset(options.dataset_entropy)
train_X4, _, test_X4, _ = load_dataset_properties(options.dataset_androdet)
nlp_X = process_trainset(train_X2, test_X2, 'int')
entropy_X = process_trainset(train_X3, test_X3, 'float')
androdet_X = process_trainset(train_X4, test_X4, 'float')
logging.info("CREATE MODEL")
model = create_model(activation=network['activation'],
optimizer=network['optimizer'],
learning_rate=network['learning_rate'],
output_size=train_Y1.shape[1],
merged_layers=network['merged_layers'])
t.get_cpu_time("PREPARATION")
logging.info("TRAIN")
try:
if options.train == 'true':
raise Exception('Force train model')
model = models.load_model(model_name)
except:
fit_one_at_time(model,
train_X1,
train_Y1,
nlp_X,
entropy_X,
androdet_X,
epochs=network['epochs'])
model.save(model_name)
t.get_cpu_time("TRAIN")
logging.info("TEST on TRAIN")
score_one_at_time(model, train_X1, train_Y1, nlp_X, entropy_X, androdet_X)
t.get_cpu_time("TEST on TRAIN")
logging.info("TEST")
score_one_at_time(model, test_X1, test_Y1, nlp_X, entropy_X, androdet_X)
t.get_cpu_time("TEST")
