Workflows for Machine Learning and Reproducible Science Laboratory
Praktikum Workflows für maschinelles Lernen und reproduzierbare Wissenschaft
6 CPirregularlyEN
Prerequisites
Recommended, as written
• Knowledge of the content of the module 14038 Computing at Scale in Machine Learning: Distributed Computing and Algorithmic Approaches • good knowledge of Python and Linux command line • basic knowledge of machine learning methods
Contents
The laboratory is implemented as one or several study projects using workflows for ML and science and related tools. For each project the following work steps have to be accomplished: • Getting acquainted with the state of the art • Project planning • Selecting the models and mechanisms to be used • Implementation of a prototype • Testing and evaluating the prototype • Documentation
Learning outcomes
After successfully completing the module, students will have acquired practical skills in developing and implementing machine learning workflows. They will be able to apply state-of-the-art-tools to utilize machine learning for reproducible science through reproducible, replicable, scalable, and portable workflows, related visualization and writing tools, and techniques relevant to implementing GDPR provisions for sensitive data.
Assessment
• Executable and tested prototype (50% of total marks) • Complete documentation, 10-20 pages (20% of total marks) • Successful intermediate presentation of results, 10 min (10% of total marks) • Successful final presentation of results, 20 min (20% of total marks) A student passed the module, if he/she achieves 75% of the total.
Literature (1)
- Depending on the projects, the relevant literature or material will be shared in the first session of the semester.
Remarks
• Study programme Artificial Intelligence M.Sc.: Compulsory elective module in complex „Seminars or Laboratories“ • Study programme Informatik M.Sc.: Compulsory elective module in complex "Seminare oder Praktika" (level 400)
Source: BTU's module description · retrieved 29 Sep 2026Original