After successful completion of the course, students are able to...
- Formulate problems as specific Machine Learning tasks
- Understand of a range of machine learning algorithms and their characteristics
- Select the fitting methods for a specific learning goal
- Explain data preprocessing techniques
- Evaluate the methods for their suitability
Principles of Supervised and Unsupervised Machine Learning, including pre-processing and Data Preparation, as well as Evaluation of Learning Systems. Machine Learning models discussed may include e.g. Decision Tree Learning, Model Selection, Naive Bayes, Bayesian Networks, Basic Regression Techniques, Support Vector Machines, Random Forests, Perceptron, Neural Networks, Deep Learning, as well as ensemble methods. The course also gives a short introduction to Automated Machine Learning and Reinforcement Learning. Didactical Concept:-Lectures-Exercises:students will compare different machine algorithms for particular data sets, and have to implement a machine learning algorithm - Presentation of algorithms by students - Discussion of reports that summarize the comparison of machine learning algorithmsAssessment: is based on written exam, report, and implemented machine learning algorithms
Preliminary talk (Vorbesprechung) & Intro: 05.03.2024, in-class
The course contains classroom lectures and exercises. Exercises include the application of machine learning techniques for various data sets and implementation of machine learning algorithms. The exercises are prepared at home and will be presented/discussed during the exercise classes.
This course support will be only in TUWEL - all lecture materials and news about the lecture will be made available there, and all questions regarding the course should be asked in the TUWEL forum *only*, not via TISS.
To get access to the TUWEL course, apply in TISS, and after confirmation, you can follow the TUWEL link above
ECTS Breakdown:
13 classes (including prepration): 34 h
3 classes for presentations/discussions (including preparation): 9
Assignments: 39.5 h
exam: 30 h
---------------
total: 112.5 h
- Solving of exercises regarding experiments in machine learning, using a software toolkit of the student's choice (e.g. Python scikit-learn, Matlab, R, WEKA, ...): 50%
- Written exam at the end of the semester: 50%
- Written exam - most likely on-line via TUWEL. If the pandemic situation allows face-to-face exams at the scheduled time, we would switch to an in-class exam.
Acceptance to the course will be by the lecturers. Priority is given to
1.) Students that have this course as a compulsory or elective course (i.e. most computer science studies)
2.) ERASMUS students that have Machine Learning in their learning agreement. You need to contact the lecturers (Rudolf Mayer, Nysret Musliu) and send them your signed learning agreement.
3.) PhD students from the Faculty of Informatics.
4.) Students that are currently in a bachelor programme of any of the studies mentioned in 1.), and are finishing their studies in the current semester. You will need to contact the lectureres (Rudolf Mayer, Nysret Musliu) and state your expected graduation, and which master programme you will continue.
5.) If there are still free places afterwards, they will be assigned to master and PhD students from other faculties, and finally to all other students from other faculties. You need to contact the lecturers (Rudolf Mayer, Nysret Musliu) and state why the course is important for your studies. Note that the registration can be confirmed only when for the registration period ends.
Self-Organising Systems (188.413) offers complementary topics in unsupervised data analysis. Information Retrieval (188.412) applies principles from Data Mining, Machine Learning
As a subsequent course, Problem Solving and Search in Artificial Intelligence (181.190) teaches some problem solving techniques that can be used in machine learning, and "Security, Privacy and Explainability in Maschine Learning" (194.055) offers topics in privacy-preserving machine learning (e.g. federated learning) and security of machine learning (e.g. adversarial attacks, model stealing, ...)
