def classify(train_dataset, train_labels, valid_dataset, valid_labels,
test_dataset, test_labels):
num_samples, feature_size = train_dataset.shape
num_labels = train_labels.shape[1]
batch_size = 80
keep_prob = 0.90
beta = 0.0000
weights = [1.0, 100.0]
num_epochs = [60]
init_lrs = [0.001]
hidden_layer_sizes = [
[400],
# [800, 640], [160, 80], [80, 40],
# [400, 360, 320],
# [160, 120, 80], [120, 80, 40],
]
for hidden_layer_size in hidden_layer_sizes:
for init_lr in init_lrs:
for num_epoch in num_epochs:
num_steps = int(train_dataset.shape[0] / batch_size *
num_epoch)
decay_steps = num_steps / num_epoch
mp_classifier = MultilayerPerceptron(feature_size,
hidden_layer_size,
num_labels,
init_lr,
decay_steps,
beta,
tf.nn.relu,
tf.nn.l2_loss,
weights,
tf.train.AdamOptimizer,
name='GAN-MLP-UNSW')
mp_classifier.train_with_labels(train_dataset, train_labels,
batch_size, num_steps,
valid_dataset, valid_labels,
test_dataset, test_labels,
keep_prob)
hyperparameter = {
'hidden_layer_size': hidden_layer_size,
'init_lr': init_lr,
'num_epochs': num_epoch,
'num_steps': num_steps,
'regularization beta': beta,
'optimizer': 'AdamOptimizer',
'keep_prob': keep_prob,
'act_func': 'RELU',
'class_weights': weights,
'batch_size': batch_size,
}
hyperparameter_summary(mp_classifier.dirname, hyperparameter)
f = open(mp_classifier.dirname + '/test.log')
print(f.read())
f.close()
mp_classifier.exit()
decay_steps = num_steps // num_epoch
mp_classifier = MultilayerPerceptron(feature_size,
hidden_layer_size,
num_labels, init_lr,
decay_steps, beta,
tf.nn.relu,
tf.nn.l2_loss, weights,
tf.train.AdamOptimizer,
name='PureMLP-UNSW2C')
mp_classifier.train_with_labels(train_dataset, train_labels,
batch_size, num_steps,
valid_dataset, valid_labels,
test_dataset, test_labels,
keep_prob)
hyperparameter = {'hidden_layer_size': hidden_layer_size,
'init_lr': init_lr,
'num_epochs': num_epoch,
'num_steps': num_steps,
'regularization beta': beta,
'optimizer': 'AdamOptimizer',
'keep_prob': keep_prob,
'act_func': 'RELU',
'class_weights': weights,
'batch_size': batch_size, }
hyperparameter_summary(mp_classifier.dirname,
hyperparameter)
f = open(mp_classifier.dirname + '/test.log')
print(f.read())
f.close()
mp_classifier.exit()
trans_func=tf.nn.sigmoid,
num_labels=2,
name=encoder_name)
rbm.train_with_labels(train_dataset, train_labels, int(num_steps),
valid_dataset, rbm_lr)
test_loss = rbm.calc_reconstruct_loss(test_dataset)
print("Testset reconstruction error: %f" % test_loss)
hyperparameter = {
'#hidden units': num_hidden_rbm,
'init_lr': rbm_lr,
'num_epochs': num_epoch,
'num_steps': num_steps,
'act_func': 'sigmoid',
'batch_size': batch_size,
}
hyperparameter_summary(rbm.dirname, hyperparameter)
hrand = np.random.random((train_dataset.shape[0], num_hidden_rbm))
rbm_train_dataset = rbm.encode_dataset(train_dataset, hrand)
print('Encoded training set', rbm_train_dataset.shape)
hrand = np.random.random((valid_dataset.shape[0], num_hidden_rbm))
rbm_valid_dataset = rbm.encode_dataset(valid_dataset, hrand)
print('Encoded valid set', rbm_valid_dataset.shape)
hrand = np.random.random((test_dataset.shape[0], num_hidden_rbm))
rbm_test_dataset = rbm.encode_dataset(test_dataset, hrand)
print('Encoded test set', rbm_test_dataset.shape)
maybe_npsave('trainset.rbm', rbm_train_dataset, True)
maybe_npsave('validset.rbm', rbm_valid_dataset, True)
maybe_npsave('testset.rbm', rbm_test_dataset, True)