Introduction in Deep Learning for Cybersecurity

The threat of cyber-attacks has long been known, essentially since the Internet became our daily routine. Like the internet itself and computer technologies that incorporate it, cyber threats are improving every year. Now artificial intelligence and machine learning work on both sides of the trenches – cybersecurity professionals and hackers. And at this moment deep learning comes into force and sets new rules in the game.

Deep learning-based methods outperform classical methods in all tasks that are too complex to describe using rules. Cybersecurity tasks often involve defending systems from human actors whose behaviour is also hard to describe with regulations. This opens interesting opportunities for cross-domain research. The interest in deep learning research has grown exponentially in recent years.

In this summer school, participants will experience an introduction in fundamental deep learning methods. Participants will be required to apply linear algebra, calculus, information theory, probability theory for cybersecurity related tasks. The course will use Python and PyTorch as a programming framework. Participants are required to be fluent in programming languages and mathematics. The mini course will include anomaly detection tasks in network traffic and breaking image-based captchas using deep learning methods. Participants will also have access to the High-Performance Cluster (HPC), also known as a supercomputer, to train models on research grade GPUs.