The knowledge gained from the pre-lab and experience from the hands-on lab in this module allows students for further implementation of Anomaly Detection in neural network systems.
VAE-GAN Hybrid Algorithm:
As explained in the pre-lab of this module, a VAE-GAN Hybrid algorithm is also good for implementing anomaly detection in a neural network. In this exercise, students are to code and demonstrate this algorithm with Google Colab.
As a refresher, VAE uses its "compressing" ability towards the typical traffic patterns as GAN makes sure that the reproductions are as "real" to the original as possible. So, when the hybrid algorithm finds anomalous data, reconstructing it would be difficult and this would be reflected when the GAN discriminator identifies the error as "fake."