Presentation is loading. Please wait.

Presentation is loading. Please wait.

Adaptive Multi-objective Differential Evolution with Stochastic Coding Strategy Wei-Ming Chen 2011.12.15.

Similar presentations


Presentation on theme: "Adaptive Multi-objective Differential Evolution with Stochastic Coding Strategy Wei-Ming Chen 2011.12.15."— Presentation transcript:

1 Adaptive Multi-objective Differential Evolution with Stochastic Coding Strategy Wei-Ming Chen 2011.12.15

2  DIFFERENTIAL EVOLUTION  “AS-MODE”  EXPERIMENTS AND COMPARISONS  CONCLUSIONS Outline

3 DIFFERENTIAL EVOLUTION

4  Location and Range AS-MODE

5  Many possible Fs and CRs  If it performs better, use it more in next generation ! AS-MODE

6  Initialization AS-MODE

7  Updating operation  Select one population  Find the neighbors  Is any one of the neighbors dominates the population ?  Yes : extend the range  No : reduce the range  Add “good neighbors” into next generation AS-MODE

8

9  Mutation, Crossover and Selection  Mutation and Crossover  Selection : the same way as NSGA-II AS-MODE

10  Update values  Range  Probabilities of candidate values AS-MODE

11  IGD : judge the quality of solution  P* : a set of solution is uniformly distributed along the Pareto front  P : the points of our solution  d(v, P) : the shortest distance between v and points in P EXPERIMENTS AND COMPARISONS

12

13

14  stochastic coding strategy  makes individuals easier detect their surrounding region  Multi mutation factor F and crossover probability CR  make populations can adjust to better algorithm  Efficiency  a little worse than NSGA-II in single generation  maybe can reduce total generation  Better ? CONCLUSIONS

15  Thank you. Q & A


Download ppt "Adaptive Multi-objective Differential Evolution with Stochastic Coding Strategy Wei-Ming Chen 2011.12.15."

Similar presentations


Ads by Google