Network intrusion detection with Deep Learning

• Network intrusions: methods of attacks and their properties
• Fundamentals of DL in network security
• Datasets
• Feature selection, extraction and processing
• DL model design and implementation
• DL model training and hyper-parameters tuning
• System deployment
• Advanced topics in ML-based intrusion detection:
  - Adversarial Machine Learning
  - Privacy-preserving collaborative learning
• Challenges of real-world intrusion detection with ML
  - Anomaly detection in Industrial Control Systems
  - Concept drift
  - The Open-world challenge

A minimum of 75% attendance is required.

Teaching methods

The course will be provided in English and combine theoretical concept lectures with live demonstrations. The theoretical lectures will start by covering cybersecurity aspects that are relevant to the course, before focusing on deep learning algorithms for cybersecurity applications.
The demonstrations will be based on the Python programming language, Tensorflow and Keras libraries and Jupyter notebooks.

Assessment methods

The examination focuses on presenting and discussing a research paper related to Intrusion and Anomaly Detection using ML techniques. Students have the autonomy to select a research paper from a list that the teacher will provide before the end of the course.

Bibliography

• Aurelien Geron. Hands-on machine learning with Scikit-Learn, Keras, and TensorFlow. ” O’Reilly Media, Inc.”, 2022.
• Clarence Chio and David Freeman. Machine Learning and Security. ” O’Reilly Media, Inc.”, 2018.