Machine Learning for Graphs and Sequential Data

What it is about

You will learn methods of machine learning for graph data and sequential data. You understand models for texts and temporal sequences as well as techniques for networks and graphs and can apply and evaluate them to real tasks. In the end you can select suitable procedures and assess their strengths/weaknesses for non-independent data.

What you will be able to do

• Understanding the data mining / machine learning process
• Application of models for sequential data (e.g., Markov models, HMMs, RNN/LSTM, Transformer)
• Application of methods for graph data (e.g., PageRank, community detection, GNNs)
• Creation and evaluation of embeddings for words, nodes and graphs
• Assessment and improvement of model robustness (including adversarial examples and certifiable robustness)
• Practically oriented implementation and critical evaluation of algorithms

What the module consists of

• LectureProvision of the theoretical foundations and methods
• problems for individual studyDeepening and independent practice of the content
• assignments including project workpractical implementation, implementation and evaluation

Teaching method

• LectureExplain concepts and models
• Problems for individual studyIndependent practice to consolidate understanding
• Assignments including project workApplying and evaluating methods in practical tasks