Lecture 2

Question 1

when is a function called unimodal?

Answer 1

f is called unimodal if it has exactly one maximum or minimum (eg. parabool)

if it has multiple (local) minima or maxima, it is called multimodal

Question 2

what does parameter fitting mean?

Answer 2

tuning the model parameters to minimize the distance between the target of a certain data point and the outcome of the model for that data point

Question 3

what are the 7 key features of Genetic Algorithms?

Answer 3

• discrete representations
• emphasis on crossover
• no self-adaptation
• larger population sizes
• probabilistic selection
• developed in the US
• theory focused on schema processing

Question 4

what are the 7 key features of Evolution Strategies?

Answer 4

• mixed-integer capabilities
• emphasis on mutation
• self-adaptation
• smaller population sizes
• deterministic selection
• developed in Europe
• theory focused on convergence speed

Question 5

what are the 7 main components of EAs?

Answer 5

• representation of individuals (coding)
• evaluation method for individuals (fitness)
• initialization procedure for first generation
• definition of variation operators (mutation, crossover)
• parent (mating) selection mechanism
• survivor (environmental) selection mechanism
• technical parameters (hyperparameters)

Question 6

what are 6 advantages of EAs?

Answer 6

• widely applicable, also in cases where no good solution techniques are available
• no presumptions w.r.t. search space
• low development costs
• solutions have straightforward interpretations
• can run interactively, always deliver solutions
• self-adaptation of strategy parameters

Question 7

what are 3 disadvantages of EAs?

Answer 7

• no guarantee for finding optimal solutions within a finite amount of time (this is true for all global optimization methods)
• no complete theoretical basis (yet), but much progress is being made
• parameter tuning sometimes based on trial and error