WiSe 23/24: Cybersecurity and AI III
Gerhard Wunder
Comments
Cybersecurity and AI is a four semester lecture cycle covering all main aspects of Cybersecurity and AI. This course teaches the fundamentals of Representation Learning and Anomaly Detection. It is in principle possible to start the lecture cycle at any semester.
The course outline is:
1. Introduction
Representation Learning (unique-, shared-, rate distortion limits)
Anomaly Detection (overview, approaches, metrics)
2. Classical ML
Types of datasets (images, text, tabular data, )
Preprocessing Tools (statistical tools, matrix factorizations, …)
Advanced Tools (Deconvolution, sparse processing, super resolution, …)
ML Toolbox (SVM / kernel methods, clustering, random /isolation forests, …)
ML-based anomaly detection and examples
3. Deep Learning I
MLP and CNN basics
Autoencoder
Reconstruction based anomaly detection
4. Deep Learning II
VAE, GAN, normalizing flows, cross-domain families (CycleGAN etc.)
Transformer I (Multi-head attention)
Transformer II (Tab/FT transformer, ViT)
Self-supervised learning
Advanced deep anomaly detection algorithms
5. Application examples in 5G/6G and IoT networks
16 Class schedule
Additional appointments
Thu, 2024-03-07 11:00 - 13:00
Location:
T9/137 Konferenzraum (Takustr. 9)
Location:
T9/137 Konferenzraum (Takustr. 9)
Regular appointments