Everyone who wants to participate please write Iris an Email, indicating your selected topic.
The list of topics can be found in the introduction slides.
Dates
will be discussed with the participants
Preliminary discussion
Tuesday 13:00 - 13:30 in Z816
Lecturers
- Michael R. Berthold
- Iris Adä
Overview
Hours per week : 2
Credits : 4
Participants: Min: 3 Max: 12 Expected: 6
Type of course: Seminar (Bachelor/Master)
Language: english
Content
In this seminar we will discuss different variations of meta learning, also known as ensemble learning.
The main assumption of meta learning is, that if one simple learner produces quite well results, the combination of multiple ones produce way better results.
We will discuss different meta learning strategies, together with possible use cases.
Possible Topics are :
- Boosting
- Random Forests
- Dynamic Weighted Majority
- Stacked Generalization
- ....
Form of Exam
oral presentation (30 minutes) and written examination(8pages) at the end of semester
Preliminaries
none
