TU Wien:Selbstorganisierende Systeme VU (Rauber)/Prüfung 2020-06-17

The exam consisted of 5 Questions. The work time was 60 minutes. Questions are not exactly like in the exam, but from memory. The main points should be there.

Question 1: SOM

10 True/False questions about SOM. Each correct answer was worth 2 points, and each false answer -1. No answer was worth 0 points (so 20 points max)

General questions about the SOM, like visualizations, magnification factors, border effect, etc. Requires good understanding of the principles of SOM training (first SOM slide set) and visualizations (second SOM slide set)

Question 2: SOM: P/U-Matrix

Describe how P- and U-Matrix visualizations are calculated and what can we learn from the respective visualizations.

Question 3: Genetic algorithms

a) What problems can be solved with Genetic algorithms

b) Name 3 genetic operators that are used in Genetic Algorithms and how they work

c) Why is diversity an issue, and how can it be controlled?

Question 4: Agents/Ant System

a) Relate the principles of self-organization to Agents and Ant Colony Optimization and where they apply and where they don't

b) Name two problems in the original ACO and two extensions that try to overcome them and how they work

Question 5: PSO

a) Name two methods of discretization in PSO, at what stages of the algorithm they are applied and their advantages and disadvantages. Also discuss if and how they can be applied to combinatorial problems (permutations)

b) Explain extensions of the neighborhood topology in the original PSO, what problem they try to overcome, how they work, and which types there are.

c) Explain the concept of stigmergy and discuss the advantages. Name two examples where stigmergy is used in swarm optimization.