def __init__(self, config, cell_type, event_file, time_file, is_training):
self.cell_type = cell_type
self.event_file = event_file
self.time_file = time_file
self.num_layers = config.num_layers
self.hidden_size = config.hidden_size
self.g_size = config.g_size
self.filter_output_dim = config.filter_output_dim
self.filter_size = config.filter_size
self.batch_size = config.batch_size
self.num_steps = config.num_steps
self.is_training = is_training
self.keep_prob = config.keep_prob
self.res_rate = config.res_rate
self.length = config.output_length
self.vocab_size = config.vocab_size
self.learning_rate = config.learning_rate
self.LAMBDA = config.LAMBDA
self.delta = config.delta
self.gamma = config.gamma
self.event_to_id = read_data.build_vocab(self.event_file)
self.train_data, self.valid_data, self.test_data = read_data.data_split(
event_file, time_file)
self.embeddings = tf.get_variable("embedding",
[self.vocab_size, self.hidden_size],
dtype=tf.float32)
self.build()
def __init__(self, config, cell_type, event_file, time_file, is_training):
self.alpha = 1.0
self.cell_type = cell_type
self.event_file = event_file
self.time_file = time_file
self.num_layers = config.num_layers
self.hidden_size = config.hidden_size
self.g_size = config.g_size
self.filter_output_dim = config.filter_output_dim
self.filter_size = config.filter_size
self.batch_size = config.batch_size
self.num_steps = config.num_steps
self.n_head, self.mh_size = 4, 50
self.n_g = 3 # config.num_gen
self.is_training = is_training
self.keep_prob = config.keep_prob
self.res_rate = config.res_rate
self.length = 20 # config.output_length
self.vocab_size = 52739 # config.vocab_size
self.learning_rate = config.learning_rate
self.lr = config.learning_rate
self.LAMBDA = config.LAMBDA
self.gamma = config.gamma
self.train_data, self.valid_data, self.test_data = read_data.data_split(
event_file, time_file, shuffle=True)
self.embeddings = tf.get_variable("embedding",
[self.vocab_size, self.hidden_size],
dtype=tf.float32)
self.sample_t = tf.placeholder(
tf.float32, [self.batch_size, self.num_steps + self.length])
self.target_t = tf.placeholder(tf.float32,
[self.batch_size, self.length])
self.inputs_t = tf.placeholder(tf.float32,
[self.batch_size, self.num_steps])
self.targets_e = tf.placeholder(tf.int64,
[self.batch_size, self.length])
self.input_e = tf.placeholder(tf.int64,
[self.batch_size, self.num_steps])
self.build()
FORMAT = "%(asctime)s - [line:%(lineno)s - %(funcName)10s() ] %(message)s"
DATA_TYPE = event_file.split('/')[-1].split('.')[0]
logging.basicConfig(filename='log/{}-{}-{}.log'.format(
'MM-CPred', DATA_TYPE, str(datetime.datetime.now())),
level=logging.INFO,
format=FORMAT)
handler = logging.StreamHandler()
handler.setFormatter(logging.Formatter(FORMAT))
logging.getLogger().addHandler(handler)
logging.info('Start {}'.format(DATA_TYPE))
# Read data
train_data, valid_data, test_data = read_data.data_split(event_file,
time_file,
shuffle=True)
# initialize the model
model = Model.MM_CPred(args)
# Get Config and Create a Session
config = tf.compat.v1.ConfigProto()
config.gpu_options.allow_growth = True
config.allow_soft_placement = True
sess = tf.compat.v1.Session(config=config)
sess.run(tf.initialize_all_variables())
# Run Train Epoches
train_i_e, train_t_e, train_i_t, train_t_t = read_data.data_iterator(
train_data, args.T, args.len, overlap=True)
import math import matplotlib.pyplot as plt from tensorflow import keras # from sklearn.model_selection import train_test_split from read_data import read_datasets, get_label, data_split, pair_batch_generator from Lin_models import Lin_Net # Load data&label cover_dir = r'.\cover_dir' stego_dir = r'.\stego_dir' X_c = read_datasets(cover_dir) X_s = read_datasets(stego_dir) # Data preprocessing X_train, X_val, X_test = data_split(X_c, X_s) y_train = get_label(X_train.shape[0], positive=0.5) y_val = get_label(X_val.shape[0], positive=0.5) y_test = get_label(X_test.shape[0], positive=0.5) X_train = X_train.reshape(-1, X_train.shape[1], 1) X_val = X_val.reshape(-1, X_val.shape[1], 1) X_test = X_test.reshape(-1, X_test.shape[1], 1) y_train = keras.utils.to_categorical(y_train, num_classes=2) y_val = keras.utils.to_categorical(y_val, num_classes=2) y_test = keras.utils.to_categorical(y_test, num_classes=2) # Build model model = Lin_Net(X_train) # Before fine-tuning the network, you should change the names of the HPF layer and Dense(2) layer.
