LFU:online

Under the successful completion of this course students understand the basics of  knowledge discovery. They can use selected methods of knowledge discovery data analysis and are capable of interpreting data and presenting it visually and verbally.

This course will give an overview of machine learning and data mining approaches for knowledge extraction from large data sets. It will cover the entire machine learning and data mining process with topics on supervised as well as unsupervised learning techniques and empirical evaluation. Learning methods covered will range from classical approaches, such as decision trees, support vector machines, and neural networks including deep learning, to selected approaches from current research.

Preliminary description of course content:

• Day 1: Introduction, Design of Knowledge Discovery Experiments, Decision Trees
• Day 2: Linear Classifiers, Non-Linear Classifiers, Kernels, and SVM
• Day 3: Neural Networks and Deep Learning, Explainable AI
• Day 4: Text Mining, Relation Extraction and Relational Learning
• Day 5: Graph Neural Networks, Repetition

The mode of the course will contain presentations of lecture slides by lecturer; delivery of hands-on exercises using analysis examples; and the presentation of exercises and mini projects by participants.
It will be first held online as a block event from August 1-5, 2022. Subsequently there will be three meetings distributed over August and September to support students in completing their projects (implementation tasks). Finally, at the last meeting the course participants will be asked to present their work which will be then discussed and evaluated.

Participants are graded through marks of written exams, mini exercises and in-class presentation of their final projects. Details will be announced in the first session.

The reading resources will be provided in the first unit.

For practical components in this course, programming skills in Python are required to achieve defined learning outcomes. Therefore, knowledge of Python and in particular familiarity with general concepts of computer programming (e.g., variables, control structures) is needed. We recommend this course to students who are familiar with Python or are able to learn it quickly thanks to their general programming skills.
Due to a compact schedule, this course is very dense. Therefore, it may take up to full-time in the first week and on average over the entire period - half-time.
The acceptance procedure is based on prioritised randomisation. Students advanced in completion of the Digital Science minor get precedence.