Lecture 1 Flashcards
What are the three main ingredients for EA?
1) Have a Populations
2) Select better Individuals
3) Perform Variation
Describe the psuedo code for general EAβs
1 π‘ β 0
2 π·π‘β πππππ‘ππ΄πππΈπ£πππ’ππ‘ππΌπππ‘ππππΌππππ£πππ’πππ π
3 πππππ π‘πππππππ‘ππππΆπππ‘ππππππππ‘πππ‘ππ ππππ(π·π‘) π
π
3.1 πΆπ‘ β πππππ‘ππ΄πππΈπ£πππ’ππ‘πππππ πππππ(π·π‘)
3.2 π·π‘+1 β π πππππ‘ππ’ππ£ππ£ππππΌππππ£πππ’πππ (π·π‘, πΆπ‘)
3.3 π‘ β π‘ + 1
What is a Black Box Optimalisation perspective?
The Black Box Optimization (BBO) perspective:
β Maximize F(x), x β S
β No prior knowledge of F
β Guess new x and evaluate
β Minimize time (or number of evaluations)
What is a White Box Optimalisation perspective?
The White Box Optimization (WBO) perspective:
β Everything known and sufficiently understood
β Use knowledge to your advantage
* Solve analytically
* Build problem-specific algorithms
Which is better on complex problems, WBO or BBO?
BBO
Which is better on simple problems, WBO or BBO?
WBO
For which case is optimum guaranteed? WBO, BBO or both?
only WBO
What is the problem with WBO compared to BBO?
Not well generalisable
Can optimum be guaranteed in BBO?
Only in infinite runs and infinite time.