Abstract | ||
---|---|---|
A DE approach based on a new measure of population diversity and a novel parameter control mechanism is proposed with the aim of introducing a good behavior of the algorithm. The ratio of the new defined population diversity of different generations is equal to that of the population variance, therefore the adaption of parameter can use some theoretical results in(19). Combining with the method in(18), we can adjust the mutation factor F and the crossover rate CR at each generation in the searching process. The performance of the proposed algorithm (DE-F&CR) is compared to the basic DE and other four DE algorithms over 25 standard numerical benchmarks provided by the IEEE Congress on Evolutionary Computation 2005 special session on real parameter optimization. The results and its statistical analysis show that the DE-F&CR generally outperforms the other algorithms in multi-modal optimization. |
Year | DOI | Venue |
---|---|---|
2013 | 10.1080/18756891.2013.816064 | INTERNATIONAL JOURNAL OF COMPUTATIONAL INTELLIGENCE SYSTEMS |
Keywords | Field | DocType |
Intelligent algorithm, Differential evolution, Population diversity, Adaptive parameter control | Mathematical optimization,Population variance,Differential evolution,Population diversity,IEEE Congress on Evolutionary Computation,Artificial intelligence,Crossover rate,Parameter control,Machine learning,Differential evolution algorithm,Mathematics,Statistical analysis | Journal |
Volume | Issue | ISSN |
6 | 6 | 1875-6891 |
Citations | PageRank | References |
1 | 0.37 | 21 |
Authors | ||
3 |