WIW34090 – Deep Reinforcement Learning

Module
Deep Reinforcement Learning
Deep Reinforcement Learning
Module number
WIW34090
Version: 1
Faculty
Business Administration
Level
Master
Duration
1 Semester
Semester
Summer and Winter semester
Module supervisor

Prof. Dr. Christian-Andreas Schumann
Christian.Schumann(at)fh-zwickau.de
Lecturer(s)

Lecturer of the International Black Sea University (IBSU)

Course language(s)

English
in "Deep Reinforcement Learning"
ECTS credits

5.00 credits

Workload

150 hours

Courses

3.00 SCH (2.00 SCH Vorlesung | 1.00 SCH Internship)
Self-study time

105.00 hours
78.75 hours Self-study - Deep Reinforcement Learning
26.25 hours Projektarbeit(en) - Deep Reinforcement Learning
Pre-examination(s)

Participation
in "Deep Reinforcement Learning"
Examination(s)

schriftliche Prüfungsleistung - Computer project
Module examination | Examination time: 90 min | Weighting: 100% | wird in englischer Sprache abgenommen
in "Deep Reinforcement Learning"
Media type
No information
Instruction content/structure
  • OpenAi Gym basics
  • Basic simulations
  • Defining principles of markov process
  • Defining of Transition matrix
  • Calculating probabilities
  • Introducing to hidden Markov Model (HMM)
  • Rewards and returns
  • Episodic and continuous tasks
  • Discount Factor
  • The Policy function
  • State-value function
  • Temporal Difference Learning(TD)
  • TD learning
  • TD prediction
  • TD control
  • Q learning
Qualification objectives

The goal of the subject is to teach students methodologies of Reinforcement learning, how  to  train  agent  to find the optimal way of achieving specific tasks in  an unknown and stochastic environment , using special programming language python and its corresponding libraries-Tens or Flow/pytorch and OpenAI Gym, students will learn how to simulate Reinforcement  learning tasks and analyze results, using different types of Deep Learning techniques different types of neural networks, modeling and simulating different algorithms will be done. Definitions of the Markov models and stochastic environment (random Environment)  will be  introduced and used in different algorithms and Policy functions.
Special admission requirements

Basic Knowledge of python programming language, mathematical thinking skill and Machine Learning basic knowledge
Recommended prerequisites
No information
Continuation options
No information
Literature
  • Hands-On Reinforcement Learning with Python, Master reinforcement and deep reinforce-ment learning using OpenAI Gym and TensorFlow, Sudharsan Ravichandiran
Notes

This subject will be studied at IBSU, Georgia Tbilissi as an elective one in the 2nd and 3rd semester.

Lecturer: Prof. Dr. Davit Datuashvili

The exam is computer-based.
Assignment to curriculum
No information
Close detail view
© Modulux (Version 4.4)