Anomaly Generation Using Generative Adversarial Networks in Host-Based Intrusion Detection, 2018

Key information

  • Dataset: ADFA-LD
  • Architecture: Cycle-GAN
  • The purpose is to detect anomalies

Information from the paper

  • Converted existing data into images and produced synthetic anomalous data using GAN to balance the dataset
  • Used MLP for classification

Possible improvement(s) or extension(s)

  • Introducing data augmentation befor CycleGAN could be interesting