hidden_layers = [800, 500, 300, ]
noise_layers = [0.6, 0.4, ]
pretr_activ = 'sigmoid'
pretr_opt = 'adadelta'
pretr_loss = 'mse'
fine_activ = 'softmax'
fine_opt = 'adadelta'
fine_loss = 'categorical_crossentropy'
clean = lambda words: [str(word)
for word in words
if type(word) is not float]
print('\n{}\n'.format('-' * 50))
print('Reading the dataset..')
preprocessor = Preprocess(**preprocess_params)
print('Spliting the dataset..')
enron_dataset = preprocessor.dataset
enron_dataset = utils.split_dataset(x=enron_dataset['body'].values,
y=enron_dataset['label'].values)
print('Transforming dataset into vectors and matrices..')
enron_dataset = preprocessor.transform(dataset=enron_dataset)
vocabulary = preprocessor.vocabulary
print('\n{}\n'.format('-' * 50))
print('Building model..')
encoders = []
noises = []
print("\n{}\n".format("-" * 50))
if CONFIG["dataset"]["generate"]:
print("Reading the dataset..")
dataset = EnronDataset(path=CONFIG["dataset"]["path"])
enron_dataset = dataset.get_dataset()
if CONFIG["dataset"]["output"]:
print("Exporting the dataset..")
dataset.to_csv(filepath=CONFIG["dataset"]["filepath"])
if CONFIG["preprocess"]["params"]["read_csv"]:
print("Reading the dataset..")
preprocessor = Preprocess(**CONFIG["preprocess"]["params"])
else:
preprocessor = Preprocess(dataset=enron_dataset, **CONFIG["preprocess"]["params"])
if CONFIG["preprocess"]["clean_dataset"]:
print("Cleaning the dataset..")
preprocessor.clean_data()
if CONFIG["preprocess"]["output_csv"]:
print("Exporting clean dataset..")
preprocessor.dataset.to_csv(CONFIG["preprocess"]["output_csv_filepath"])
print("Spliting the dataset..")
enron_dataset = preprocessor.dataset
enron_dataset = utils.split_dataset(x=enron_dataset["body"].values, y=enron_dataset["label"].values)
