Differential Evolution Using a Neighborhood-Based Mutation Operator
read more
Citations
Differential Evolution: A Survey of the State-of-the-Art
Recent advances in differential evolution – An updated survey
Differential Evolution With Composite Trial Vector Generation Strategies and Control Parameters
Differential evolution algorithm with ensemble of parameters and mutation strategies
Recent advances in differential evolution: a survey and experimental analysis
References
Particle swarm optimization
Adaptation in natural and artificial systems
Differential Evolution – A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces
Introduction to Algorithms
Related Papers (5)
Differential Evolution – A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces
JADE: Adaptive Differential Evolution With Optional External Archive
Frequently Asked Questions (13)
Q2. What future works have the authors mentioned in the paper "Differential evolution using a neighborhood-based mutation operator" ?
In addition, the performance of the competitor DE variants may also be improved by blending other mutation strategies with judicious parameter tuning, a topic of future research. Future research may focus on providing some empirical or theoretical guidelines for selecting the neighborhood size over different types of optimization problems. It would be interesting to study the performance of the DEGL family when the various control parameters ( N P, F, and Cr ) are self-adapted following the ideas presented in [ 22 ], [ 28 ]. The conclusion the authors can draw at this point is that DE with the suggested modifications can serve as an attractive alternative for optimizing a wide variety of objective functions.
Q3. What is the DE mutation for a neighborhood?
The neighborhood-based DE mutation, equipped with selfadaptive weight factor, attempts to make a balanced use of the exploration and exploitation abilities of the search mechanism and is therefore more likely to avoid false or premature convergence in many cases.
Q4. What is the neighborhood topology for DE?
Although various neighborhood topologies (like star, wheel, pyramid, 4-clusters, and circular) have been proposed in the literature for the PSO algorithms [42], after some initial experimentation over numerical benchmarks, the authors find that in the case of DE (where the population size is usually larger than in the case of PSO) the circular or ring topology provides best performance compared to other salient neighborhood structures.
Q5. Why are the vector indices sorted randomly?
The vector indices are sorted only randomly (as obtained during initialization) in order to preserve the diversity of each neighborhood.
Q6. What is the key to the success of the search-dynamics of DEGL?
This suggests that a judicious tradeoff between the explorative and the exploitative mutation operators is the key to the success of the search-dynamics of DEGL.
Q7. What is the individual vector for a minimization problem?
G is the best individual vector with the best fitness (i.e., lowest objective function value for a minimization problem) in the population at generation G. Note that some of the strategies for creating the donor vector may be mutated recombinants, for example, (5) listed above, basically mutates a two-vector recombinant: Xi,G +F ·( Xbest,G − Xi,G).
Q8. How do the authors combine the local and global donor vectors?
Now the authors combine the local and global donor vectors using a scalar weight w ∈ (0, 1) to form the actual donor vector of the proposed algorithmVi,G = w. gi,G + (1 − w).
Q9. What are the results of the final optimization of multimodal functions?
For multimodal functions, the final results are much more important since they reflect an algorithm’s ability of escaping from poor local optima and locating a good near-global optimum.
Q10. What is the weighted weight for robust search?
Usually in the community of stochastic search algorithms, robust search is weighted over the highest possible convergence rate [56], [57].
Q11. What is the advantage of measuring the runtime complexity by counting the number of FEs?
The advantage of measuring the runtime complexity by counting the number of FEs is that the correspondence between this measure and the processor time becomes stronger as the function complexity increases.
Q12. What is the arithmetic crossover operator used to generate the trial vector Ui,?
To overcome this limitation, a new trial vector generation strategy “DE/current-to-rand/1” is proposed in [19], which replaces the crossover operator prescribed in (9) with the rotationally invariant arithmetic crossover operator to generate the trial vector Ui,G by linearly combining the target vector Xi,G and the corresponding donor vector Vi,G as follows:Ui,G = Xi,G + K · ( Vi,G − Xi,G).
Q13. How many traditional numerical benchmarks have been used to evaluate the performance of the new DE variant?
The authors have used a test bed of 21 traditional numerical benchmarks (Table IV) [47] and three composition functions from the benchmark problems suggested in CEC 2005 [48] to evaluate the performance of the new DE variant.