Fernandes, Merelo, Ramos, Rosa, presented at the Self-* workshop within the PPSN conference, Kraków 2010

1. PPSN’10 - Krakow Fernandes, Laredo, MereloandRosa A Self-Organized Criticality Online Adjustment of Genetic Algorithms’ Mutation Rate Carlos M. Fernandes1,2 J.L.J. Laredo1 J.J. Merelo1 Agostinho C. Rosa2 1Department of Architecture and Computer Technology, University of Granada, Spain 2LaSEEB-ISR-IST, Technical Univ. of Lisbon (IST), Portugal

3. In DOPs, the fitness function and the constraints of the problem are not constant. When changes occur, the solutions already found may be no longer valuable and the process must engage in a new search effort.

4. DOPs require diversity (when using population-based heuristics).

6. Hypothesis: self-organized criticality

7. Sand pile Mutation Operator

9. Evolutionary approaches to dynamic optimization

10. Self-organized criticality

11. Sand Pile Mutation Operator

12. Test Set and Results

13. Mutation Rates and distribution

15. Adaptive: the values change variation depends indirectly on the problem and the search stage

17. Memory

18. Multi-Population

20. Unlike many physical systems, SOC systems are able to self-tune to the critical point.

21. SOC has been used in EC, but there are few studies

22. Krink et al.: power-law is computed offline

23. SORIGA: uses a SOC model to introduce random immigrants in the population

25. Fernandes, Merelo, Ramos and Rosa – “Sandpile Mutation GA” PPSN’10 - Krakow TheSand Pile Mutation Grains are dropped at a rate g Mutates if a random value (0,1.0) is above the normalized fitness

27. By using a binary mask, dynamic environments are created by applying the mask to each solution before its evaluation.

28. Severity of change is controlled by setting the number of 1’s in the mask.

29. Speed of change is controlled by defining the number of generations between the application of a different mask.Severity of change: This criterion establishes how strongly the problem is changing Frequency of change: This criterion establishes how often the environment changes

31. 4-trap, knapsack and royal road

32. Compared a generational GA with Sand Pile Mutation (GGASM ) with:

33. Standard Generational GA (GGA)

34. Self-Organized Criticality Random Immigrants GA (SORIGA).

36. Uniform crossover

37. pc = 1.0

38. Binary tournament

39. Speed was set to 1200, 2400, 24000, 48000 evaluations

40. Population size n= 30, 60, 120

41. pm : 1/(16×l), 1/(8×l), 1/(4×l), 1/(2×l), 1/l, 2/l, 4/l

43. Fernandes, Laredo, Mereloand Rosa PPSN’10 - Krakow . Mutation rates anddistribution Mutation rate: m(i,j) = 1if the gene of the chromosome has mutated, and 0 otherwise n is the population size and l is the chromosome length Order-4 traps. Mutation rate median values.

44. Fernandes, Laredo, Mereloand Rosa PPSN’10 - Krakow Mutation rates anddistribution Order- dynamic trap problems. GGASM online mutation rate. Logarithm of the mutation rates abundance plotted against their values.

46. It clearly outperforms SORIGA, and it at least competitive with EIGA.

48. Study the mutation rates and mutation distribution.

49. Constrained dynamic optimization.

50. Stationary Optimization.