Introduction to Reinforcement Learning for Robotics (Summer School)
Academic Study Board of the Faculty of Engineering
Teaching language: English
EKA: T540027102
Censorship: Second examiner: None
Grading: Pass/Fail
Offered in: Odense
Offered in: Spring
Level: Bachelor
Course ID: T540027101
ECTS value: 5
Date of Approval: 14-05-2019
Duration: Intensive course
Version: Archive
Course ID
Course Title
ECTS value
5
Internal Course Code
Responsible study board
Administrative Unit
Date of Approval
Course Responsible
Programme Secretary
Offered in
Level
Offered in
Duration
Mandatory prerequisites
Learning objectives - Knowledge
1) Understand the types of learning problems that can appear in a robotic context
2) Describe key concepts, such as decision processes, value, and policy in the broad context of Reinforcement Learning
3) Identify robot learning problems as planning or control problems
4) Understand the limitations
Learning objectives - Skills
1) Analyze and select appropriate Reinforcement Learning techniques to solve robotic problems
2) Formulate adequate solutions to Reinforcement Learning problems
Learning objectives - Competences
Solve complex robotics problems using Reinforcemente Learning techniques
Content
Content - Key areas:
Introduction to Reinforcement Learning
- Differences between supervised, unsupervised and reinforcement learning.
- Decision Processes
Reinforcement Learning for Planning
- Markov Decision processes, Policies and Value functions
- Policy Iteration and Value Iteration
- Temporal-Difference Learning
Introduction to supervised learning for regression
- Regression problems
- Artificial Neural Networks
Policy Search
- Reinforcement Learning for control
- Algorithms to Optimize the return
- 3Gradient estimation methods in Reinforcement Learning
Time of classes
2 weeks in August
URL for Skemaplan
Teaching Method
Number of lessons
hours per semester
Teaching language
Examination regulations
Exam regulations
Name
Exam regulations
Examination is held
At the end of the course
Tests
Exam
EKA
T540027102
Name
Exam
Description
The examination is based on an overall assessment of:
- Attendance (80 %)
- Oral exam
Form of examination
Oral examination
Censorship
Second examiner: None
Grading
Pass/Fail
Identification
Student Identification Card
Language
English
ECTS value
5