Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.
self.generator_input_shape)
self.critic_x = self._build_model(hyperparameters, self.layers_critic_x,
self.critic_x_input_shape)
self.critic_z = self._build_model(hyperparameters, self.layers_critic_z,
self.critic_z_input_shape)
self.generator.trainable = False
self.encoder.trainable = False
z = Input(shape=(self.latent_dim, 1))
x = Input(shape=self.shape)
x_ = self.generator(z)
z_ = self.encoder(x)
fake_x = self.critic_x(x_)
valid_x = self.critic_x(x)
interpolated_x = RandomWeightedAverage()([x, x_])
validity_interpolated_x = self.critic_x(interpolated_x)
partial_gp_loss_x = partial(self._gradient_penalty_loss, averaged_samples=interpolated_x)
partial_gp_loss_x.__name__ = 'gradient_penalty'
self.critic_x_model = Model(inputs=[x, z], outputs=[valid_x, fake_x,
validity_interpolated_x])
self.critic_x_model.compile(loss=[self._wasserstein_loss, self._wasserstein_loss,
partial_gp_loss_x], optimizer=self.optimizer,
loss_weights=[1, 1, 5])
fake_z = self.critic_z(z_)
valid_z = self.critic_z(z)
interpolated_z = RandomWeightedAverage()([z, z_])
validity_interpolated_z = self.critic_z(interpolated_z)
partial_gp_loss_z = partial(self._gradient_penalty_loss, averaged_samples=interpolated_z)
partial_gp_loss_z.__name__ = 'gradient_penalty'
fake_x = self.critic_x(x_)
valid_x = self.critic_x(x)
interpolated_x = RandomWeightedAverage()([x, x_])
validity_interpolated_x = self.critic_x(interpolated_x)
partial_gp_loss_x = partial(self._gradient_penalty_loss, averaged_samples=interpolated_x)
partial_gp_loss_x.__name__ = 'gradient_penalty'
self.critic_x_model = Model(inputs=[x, z], outputs=[valid_x, fake_x,
validity_interpolated_x])
self.critic_x_model.compile(loss=[self._wasserstein_loss, self._wasserstein_loss,
partial_gp_loss_x], optimizer=self.optimizer,
loss_weights=[1, 1, 5])
fake_z = self.critic_z(z_)
valid_z = self.critic_z(z)
interpolated_z = RandomWeightedAverage()([z, z_])
validity_interpolated_z = self.critic_z(interpolated_z)
partial_gp_loss_z = partial(self._gradient_penalty_loss, averaged_samples=interpolated_z)
partial_gp_loss_z.__name__ = 'gradient_penalty'
self.critic_z_model = Model(inputs=[x, z], outputs=[valid_z, fake_z,
validity_interpolated_z])
self.critic_z_model.compile(loss=[self._wasserstein_loss, self._wasserstein_loss,
partial_gp_loss_z], optimizer=self.optimizer,
loss_weights=[1, 1, 10])
self.critic_x.trainable = False
self.critic_z.trainable = False
self.generator.trainable = True
self.encoder.trainable = True
z_gen = Input(shape=(self.latent_dim, 1))
x_gen_ = self.generator(z_gen)