icec2010 presentation

Post on 21-Jun-2015

339 Views

Category:

Technology

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

INVESTIGATING REPLACEMENT STRATEGIES

FOR THE

ADAPTIVE DISSORTATIVE MATING GENETIC

ALGORITHMCarlos Fernandes1,2

J.J. Merelo1

Agostinho C. Rosa2

1Department of Architecture and Computer Technology, University of Granada, Spain 2 L aSEEB-ISR-IST, Technical Univ. of Lisbon (IST), Portugal

SUMMARY

ADMGA

Non-Stationary Fitness Landscapes

Motivation

Replacement Strategies

Results

Conclusions and Future Work

Dissortative MatingDissortative Mating

Mating between dissimilar individuals. Higher diversity.

Disruptive effect

High selective pressure + high disruption effectparent

parent

Chromosomes are alowed to crossover if and only their Hamming Distance is above the threshold value.

The threshold self-adapts its initial value, and varies during the run according to the population diversity

1111111111111111

1111111100001111

Hamming dist.: 4selection

ADMGA differs from the SGA at the recombination stage

4

the number of positions at which the corresponding symbols are different

Adaptive Dissortative Mating Adaptive Dissortative Mating GA (ADMGA)GA (ADMGA)

ADMGAADMGAPopulation

New population = Offspring population + best parents

Selects two and computes h.d.

if h. d. > ts

if h. d. ≤ ts

Crossover and mutate

after n/2 (n is the population size)

Updates threshold

if (failed matings > successful matings) ts← ts−1else ts ← ts+1

5

diversity is controlling the threshold

population-wide elitism (or steady-state)

Stationary Fitness Functions:Stationary Fitness Functions:Scalability with Trap FunctionsScalability with Trap Functions

order-2 (k = 2) order-3 order-4

6

non-deceptive nearly-deceptive fully deceptive

Scalability with problem size

Alternative Replacement StrategiesThreshold ValueThreshold Value

Initial threshold value

n = 10,000; l = 10

n = 10; l = 10,000n = 100

order-2

Dynamic Optimization Dynamic Optimization ProblemsProblems

8

ADMGA: Dynamic ADMGA: Dynamic Optimization ProblemsOptimization Problems

Better performance on “slower” dynamic problems

The performance degrades as the optimum moves faster

9

MotivationMotivation

Improve ADMGA’s performance on faster problems

Is population-wide elitism a good or bad strategy for fast dynamic problems?

10

Replacement StrategiesReplacement Strategies

RS 1: Original

RS 2: Mutated copies of the old solutions

RS 3: Mutated copies of the best solution

RS 4: Random Immigrants (random solutions)

11

ADMGA: Dynamic ADMGA: Dynamic Optimization ProblemsOptimization Problems

Yang’s (2003) dynamic problem generator:• frequency of change (1/ε)• severity (ρ)

12

ε : 600, 1200, 2400, 4800, 9600, 19200, 38400 ρ : random

Offline performance: average of the best fitness throughout the run

Statistical tests

TestsTests

Several mutation probability and population size values.• mutation: dissortative mating affects optimal

probability• population size: avoid extra computational effort

binary tournament 2-elitism uniform crossover (p=1.0)

• Balance disruptive effect and selective pressure

13

Results TestsTestsRS 1 vs GGA

RS 2 vs GGA

ε→ 600 1200 2400 4800 9600 19200 38400

onemax − − ≈ ≈ ≈ ≈ ≈trap ≈ ≈ + + + + +

knapsack − − ≈ ≈ ≈ + +

ε→ 600 1200 2400 4800 9600 19200 38400

onemax − − − − ≈ ≈ ≈trap − − − ≈ + + +

knapsack − − − − − ≈ +

Results TestsTests

RS 2 vs EIGA

ε→ 600 1200 2400 4800 9600 19200 38400

onemax − − − ≈ ≈ ≈ ≈trap ≈ ≈ + + + + +

knapsack − − ≈ ≈ ≈ ≈ ≈

Genetic Diverstiy

Conclusions and Future Work

Mutating old solutions speeds up AMDGA on dynamic problems

Only two parameters need to be adjusted: population size and mutation rate

ADMGA is at least competitive with EIGA

Performance according to severity

Constrained Dynamic Problems

top related