AI-and-Cybersecurity-Reports

AI & Cybersecurity laboratory activity reports for the AI & Cybersecurity course of Politecnico di Torino

Authors

Giorgia Moscato, Angelo Barbera, Alessandro Genova.

Abstract

These reports aims to explore Artificial Intelligence techniques applied to Cybersecurity. Each laboratory is focused on a specific AI aspects involved in addressing cybersecurity challenges:

• Lab 1 - Introduction to Deep Learning:
  • Data preprocessing.
  • Shallow and deep neural network.
  • Impact of batch size, activation function and optimizer.
  • Overfitting and regularization.
• Lab 2 - Model Engineering:
  • Bag of word approach.
  • Embeddings.
  • Feed forward neural network.
  • Recursive Neural Network.
  • Graph Neural Network.
• Lab 3 - Natural Language Processing:
  • Tokenization.
  • BERT and Unixcoder training from scratch.
  • BERT and Unixcoder fine-tuning.
  • BERT and Unixcoder inference.
• Lab 4 - Anomaly Detection:
  • Data analysis.
  • One-Class SVM supervised and unsupervised detection.
  • Deep auto-encoder detection.
  • Auto-encoder and One-Class SVM detection.
  • Principal component analysis.
  • K-means clustering.
  • DB-Scan clustering.
  • t-SNE visualization.