def fit(self, X, y=None):
print ("fit X: ", X.shape)
print ("fit y: ", y.shape if y is not None else '')
"""
Split train and test set
"""
train_datas = self.normal_datas[X]
valid_datas = np.vstack([np.delete(self.normal_datas,X,axis=0),
self.abnormal_datas])
valid_labels = np.hstack([np.zeros(self.normal_datas.shape[0] - X.shape[0]),
np.ones(self.abnormal_datas.shape[0])])
"""
Rebuild VAE and train
"""
global global_best_test_score
tf.reset_default_graph()
# rebuild VAE
vae = VAE(**self.vae_params)
#cPickle.dump(vae, open(os.path.join(self.save_candidate_folder, 'vae_class.pkl'), 'wb'))
vae.build()
"""
Normalization
"""
vae.build_normalize(train_data=train_datas)
norm_train_datas = vae.transform_raw_data(raw_data=train_datas)
norm_valid_datas = vae.transform_raw_data(raw_data=valid_datas)
"""
Mini Batchs
"""
mini_batchs = [norm_train_datas[i:min(i + batch_size, len(norm_train_datas))] for i in
range(0, len(norm_train_datas), batch_size)]
"""
Train
"""
self.best_test_score, _ = train_vae.train(vae=vae, mini_batchs=mini_batchs, valida_data=norm_valid_datas,
valida_label=valid_labels, save_out_model=None, n_epoch=30)
# self.best_test_score, _ = train_vae.train(vae=vae, mini_batchs=mini_batchs, test_data=norm_valid_datas,
# test_label=valid_labels,save_out_model=os.path.join(self.save_candidate_folder, 'vae_tensor.ckpt'))
"""
Save result
"""
print ("Perform training with the below parameters: ")
print ("------------------------------------------- ")
print (json.dumps(self.vae_params,indent=2))
print ("------------------------------------------- ")
print ("Result (F1): ", self.best_test_score)
rs = RandomizedSearchCV(estimator=vae_hyper, param_distributions=params, n_jobs=1, verbose=2, n_iter=10,
cv=5, return_train_score=False, refit=False)
rs.fit(X=np.arange(normal_datas.shape[0]),y=None)
print ("Best parameters found: ")
print (json.dumps(rs.best_params_, indent=2))
"""
Create new VAE model for whole dataset with best parameter found
"""
tf.reset_default_graph()
vae = VAE(**rs.best_params_)
# save class instance by using cPickle, main purpose is to save parameters too.
cPickle.dump(vae, open(os.path.join(save_vae_hyper_folder, 'vae_class.pkl'), 'wb'))
vae.build()
"""
Prepare data
"""
datas = np.vstack([normal_datas, bearing_datas, gear_datas])
labels = np.hstack([np.zeros(normal_datas.shape[0]), # 0 for inlier, 1 for outlier
np.ones(bearing_datas.shape[0]),
np.ones(gear_datas.shape[0])])
train_datas, test_datas, train_labels, test_labels = utils.split_train_test(datas=datas, labels=labels, frac=0.8)
"""
Mini-batchs & perform MinMaxScaler
"""
vae.build_normalize(train_data=train_datas) # 1