Practical Course Python and Reinforcement Learning

WS   SWS: 4   ECTS: 4 WS   Examination: written (English)

 

Description

The practical course "Python and Reinforcement Learning" aims to impart basic skills and abilities in the application-oriented handling of machine learning methods, especially reinforcement learn-ing, the modeling of control processes, and Python programming. For this purpose, basic Python knowledge of syntax and common Python frameworks from data analysis and machine learning will be taught or refreshed firstly. It is followed by an introduction to the basic concepts, methods, and algorithms of machine learning, especially reinforcement learning and modeling. The ac-quired background knowledge is then transferred to application-oriented problems from control engineering and process control by modeling, implementing, and solving solutions based on rein-forcement learning for various complex use cases. Finally, the methods are applied to implement control tasks on the demonstrator "Pick and Place Unit - Demonstrator for Evolution in Industrial Plant Automation" from the chair AIS. https://www.mw.tum.de/ais/forschung/demonstratoren/ppu/.

After participating in the practical course, students will be able to

  • understand the basic theoretical principles of machine learning, specifically reinforcement learning, from an application-oriented perspective.
  • use Python and common Python frameworks for data analysis and machine learning, and to design and implement suitable pipelines for data analysis independently. Students will learn to read and understand Python-specific online documentation on open source code, analyze the code behind it, and apply common Python frameworks (e.g., Tensorflow, Keras, scikit-learn). Computational thinking will be strengthened.
  • understand complex control problems, Big Data, and Machine Learning problems in an appropriate representation form (flowchart, state diagrams) and implement them in Py-thon.
  • model and develop a control task for an automation system as a reinforcement learning task for a given problem.
  • assess and critically evaluate the use, effort, and challenges of reinforcement learning for decision making and optimization of automation engineering tasks.

Dates

Registration can only be done on TUMonline.

Contact