precision = 0.8727372840525844
recall = 0.9880776124055171
Accuracy = 0.9111958836053939
F1 = 0.9268328338571742
ROC_score = 0.953
Notice: When load the model, discriminator and bigan need to be complied again due to the use of tensorflow.optimizer.
from tensorflow.keras.optimizers import Adam optimizer = Adam(0.001, 0.5)
from keras.models import load_model
encoder = load_model("BiGAN_encoder.hdf5")
bigan = load_model("BiGAN_bigan.hdf5")
discriminator = load_model("BiGAN_discriminator.hdf5")
discriminator.compile(loss=['binary_crossentropy'], optimizer=optimizer, metrics=['accuracy'])
bigan.compile(loss=['binary_crossentropy', 'binary_crossentropy'], optimizer=optimizer)