Hubbry Logo
Genetic operatorGenetic operatorMain
Open search
Genetic operator
Community hub
Genetic operator
logo
8 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
Genetic operator
Genetic operator
Genetic operator
View on Wikipedia
from Wikipedia
Part of a series on the
Evolutionary algorithm
Genetic algorithm (GA)
Genetic programming (GP)
Differential evolution
Evolution strategy
Evolutionary programming
Related topics

A genetic operator is an operator used in evolutionary algorithms (EA) to guide the algorithm towards a solution to a given problem. There are three main types of operators (mutation, crossover and selection), which must work in conjunction with one another in order for the algorithm to be successful.[1] Genetic operators are used to create and maintain genetic diversity (mutation operator), combine existing solutions (also known as chromosomes) into new solutions (crossover) and select between solutions (selection).[2][3]

The classic representatives of evolutionary algorithms include genetic algorithms, evolution strategies, genetic programming and evolutionary programming. In his book discussing the use of genetic programming for the optimization of complex problems, computer scientist John Koza has also identified an 'inversion' or 'permutation' operator; however, the effectiveness of this operator has never been conclusively demonstrated and this operator is rarely discussed in the field of genetic programming.[4][5] For combinatorial problems, however, these and other operators tailored to permutations are frequently used by other EAs.[6][7]

Mutation (or mutation-like) operators are said to be unary operators, as they only operate on one chromosome at a time. In contrast, crossover operators are said to be binary operators, as they operate on two chromosomes at a time, combining two existing chromosomes into one new chromosome.[8][9]

Operators

[edit]

Genetic variation is a necessity for the process of evolution. Genetic operators used in evolutionary algorithms are analogous to those in the natural world: survival of the fittest, or selection; reproduction (crossover, also called recombination); and mutation.

Selection

[edit]
Main article: Selection (evolutionary algorithm)

Selection operators give preference to better candidate solutions (chromosomes), allowing them to pass on their 'genes' to the next generation (iteration) of the algorithm. The best solutions are determined using some form of objective function (also known as a 'fitness function' in evolutionary algorithms), before being passed to the crossover operator. Different methods for choosing the best solutions exist, for example, fitness proportionate selection and tournament selection.[10] A further or the same selection operator is used to determine the individuals for being selected to form the next parental generation. The selection operator may also ensure that the best solution(s) from the current generation always become(s) a member of the next generation without being altered;[11] this is known as elitism or elitist selection.[2][12][13]

Crossover

[edit]
Main article: Crossover (evolutionary algorithm)

Crossover is the process of taking more than one parent solutions (chromosomes) and producing a child solution from them. By recombining portions of good solutions, the evolutionary algorithm is more likely to create a better solution.[2] As with selection, there are a number of different methods for combining the parent solutions, including the edge recombination operator (ERO) and the 'cut and splice crossover' and 'uniform crossover' methods. The crossover method is often chosen to closely match the chromosome's representation of the solution; this may become particularly important when variables are grouped together as building blocks, which might be disrupted by a non-respectful crossover operator. Similarly, crossover methods may be particularly suited to certain problems; the ERO is considered a good option for solving the travelling salesman problem.[14]

Mutation

[edit]
Main article: Mutation (evolutionary algorithm)

The mutation operator encourages genetic diversity amongst solutions and attempts to prevent the evolutionary algorithm converging to a local minimum by stopping the solutions becoming too close to one another. In mutating the current pool of solutions, a given solution may change between slightly and entirely from the previous solution.[15] By mutating the solutions, an evolutionary algorithm can reach an improved solution solely through the mutation operator.[2] Again, different methods of mutation may be used; these range from a simple bit mutation (flipping random bits in a binary string chromosome with some low probability) to more complex mutation methods in which genes in the solution are changed, for example by adding a random value from the Gaussian distribution to the current gene value. As with the crossover operator, the mutation method is usually chosen to match the representation of the solution within the chromosome.[15][3]

Combining operators

[edit]

While each operator acts to improve the solutions produced by the evolutionary algorithm working individually, the operators must work in conjunction with each other for the algorithm to be successful in finding a good solution.[3] Using the selection operator on its own will tend to fill the solution population with copies of the best solution from the population. If the selection and crossover operators are used without the mutation operator, the algorithm will tend to converge to a local minimum, that is, a good but sub-optimal solution to the problem. Using the mutation operator on its own leads to a random walk through the search space. Only by using all three operators together can the evolutionary algorithm become a noise-tolerant global search algorithm, yielding good solutions to the problem at hand.[2]

References

[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

Genetic operator

View on Grokipedia
from Grokipedia
A genetic operator is a computational mechanism employed in genetic algorithms (GAs) and broader (EC) frameworks, such as evolutionary strategies and , to modify and evolve a population of candidate solutions. These operators simulate and genetic variation to converge on optimal or near-optimal results for complex optimization problems. They draw inspiration from biological evolution, as pioneered by John H. Holland in his 1975 work Adaptation in Natural and Artificial Systems, which formalized the core processes of GAs. The three primary genetic operators—selection, crossover, and mutation—work in tandem to drive the evolutionary process. Selection favors individuals with higher fitness; crossover exchanges genetic material between parents to produce offspring; and mutation introduces random alterations to maintain diversity. Detailed methods vary across EC paradigms. As of , genetic operators have been extended in hybrid algorithms, parallel implementations, and integrations with techniques like large language models, addressing challenges in fields such as engineering design, scheduling, and bioinformatics. Their effectiveness depends on parameter tuning, such as crossover rates (often 0.6–0.9) and low rates (e.g., 0.001). While excelling in non-differentiable, multimodal optimization, GAs can be computationally intensive, prompting ongoing research into adaptive and quantum-inspired variants.

Fundamentals

Definition and Purpose

Genetic operators are computational procedures employed in genetic algorithms (GAs) and broader frameworks to manipulate a of candidate solutions, referred to as individuals, thereby generating successive generations of solutions modeled after biological .http://www.scholarpedia.org/article/Genetic_algorithms These operators emulate natural processes by encoding solutions as chromosomes—typically in forms like binary strings or real-valued vectors—and applying transformations to evolve the population toward improved outcomes.https://link.springer.com/article/10.1007/s11042-020-10139-6 The primary purpose of genetic operators is to facilitate efficient of complex search spaces, where promoting fitter individuals drives convergence to high-quality solutions while preserving diversity to avoid premature stagnation.https://link.springer.com/article/10.1007/s11042-020-10139-6 This involves balancing exploration, which generates novel variations to broadly sample the solution space, and exploitation, which intensifies focus on promising regions to refine solutions—a dynamic essential for tackling optimization problems with multiple local .https://mitpress.mit.edu/9780262581110/adaptation-in-natural-and-artificial-systems/ Effective operator application thus enhances the algorithm's ability to approximate global in non-linear, high-dimensional domains.http://www.scholarpedia.org/article/Genetic_algorithms Central to their function are prerequisites like population representation, which structures individuals for manipulation (e.g., binary encodings for discrete problems or real vectors for continuous ones), and fitness evaluation, a scalar function that quantifies each individual's performance relative to the optimization objective.https://link.springer.com/article/10.1007/s11042-020-10139-6 Without these, operators cannot selectively propagate advantageous traits across generations.https://mitpress.mit.edu/9780262581110/adaptation-in-natural-and-artificial-systems/ The core operators—selection, crossover, and —collectively orchestrate this process.http://www.scholarpedia.org/article/Genetic_algorithms In practice, consider a GA for function optimization, where an initial random population of parameter vectors undergoes operator-induced transformations over iterations, progressively converging to near-optimal configurations that minimize or maximize the target function.https://link.springer.com/article/10.1007/s11042-020-10139-6 This evolutionary progression mirrors natural , yielding robust solutions for applications like parameter tuning in engineering design.http://www.scholarpedia.org/article/Genetic_algorithms

Historical Context

The foundations of genetic operators trace back to mid-20th-century , where concepts of self-reproduction and adaptive systems laid groundwork for . In the , explored self-reproducing automata as theoretical models for reliable computation and biological replication, influencing later ideas of algorithmic evolution through mechanisms that could generate and modify structures autonomously. Building on this, Richard M. Friedberg's 1958 work introduced via evolutionary processes, using random mutations and selection to evolve programs on an computer, marking an early empirical attempt to simulate adaptive improvement without explicit programming. In the , parallel developments in strategies (ES) by Ingo Rechenberg and Hans-Paul Schwefel at the introduced mutation-based variation and selection for continuous parameter optimization, laying foundational principles for real-valued representations in evolutionary algorithms. The formal inception of genetic operators occurred with John 's seminal 1975 book Adaptation in Natural and Artificial Systems, which introduced genetic algorithms (GAs) as computational methods inspired by natural , incorporating operators such as selection, crossover, and to mimic biological and solve optimization problems. , recognized as the founder of GAs, emphasized these operators as key to processing and building block hypotheses for adaptive search. Concurrently, Kenneth De Jong's 1975 dissertation analyzed GA behavior and parameters, including operator rates, providing empirical validation for their role in function optimization. During the 1980s, GAs and their operators gained popularity for tackling complex optimization challenges, such as the traveling salesman problem, due to advances in computing power and broader accessibility of Holland's ideas. David Goldberg's 1989 book Genetic Algorithms in Search, Optimization, and further standardized operator usage, offering theoretical and practical frameworks that propelled their adoption in and AI applications. By the 1990s, the field evolved from binary representations to real-coded GAs, enabling more precise handling of continuous parameters, as demonstrated in early works like Eshelman and Schaffer's interval-schemata analysis. This shift coincided with John Koza's 1992 integration of operators into , extending them to evolve tree-based structures for automatic .

Core Operators

Selection

Selection is a genetic operator in evolutionary algorithms that probabilistically chooses individuals from the current to serve as parents for the next generation, with selection probabilities typically proportional to their fitness values, thereby mimicking by favoring the survival and reproduction of fitter solutions. This process forms a mating pool that drives evolutionary progress toward higher-quality solutions while balancing exploration and exploitation in the search space. Common types of selection include , also known as roulette wheel selection, where the probability of selecting individual ii is given by pi=fi∑jfjp_i = \frac{f_i}{\sum_j f_j}, with fif_i denoting the fitness of individual ii. Rank-based selection addresses issues in fitness proportionate methods by assigning selection probabilities based on an individual's rank in the sorted population rather than raw fitness values, helping to prevent premature convergence caused by highly fit individuals dominating the pool. Tournament selection, another prevalent method, involves randomly sampling kk individuals (where kk is typically small, such as 2 or 3) and selecting the one with the highest fitness as a parent, offering tunable selection pressure through the choice of kk. Key parameters in selection include selection intensity, which controls the bias toward fitter individuals, and , a strategy that guarantees the preservation of the top-ranked individuals into the next generation to maintain progress. However, high selection pressure can lead to loss of population diversity, resulting in stagnation or premature convergence to suboptimal solutions, as overly aggressive favoritism reduces the of novel genotypes. The mathematical foundation of selection often revolves around the expected number of for an , calculated as E=fifˉE = \frac{f_i}{\bar{f}}, where fˉ\bar{f} is the mean fitness, reflecting the reproductive advantage of superior fitness. For linear ranking selection, a specific form of rank-based method, the selection probability is pi=1−cN+2cri−1N(N−1)p_i = \frac{1 - c}{N} + 2c \frac{r_i - 1}{N(N-1)}, where NN is the , rir_i is the rank of ii (with 1 for the best), and cc (0 ≤ c ≤ 1) is the selection intensity parameter that adjusts the pressure. In the 0/1 , where the goal is to maximize value without exceeding weight capacity, tournament selection with k=2k=2 effectively identifies and propagates individuals representing high-value, low-weight item combinations by repeatedly choosing the fitter solution from random pairs, leading to efficient convergence in benchmark instances. Selected parents subsequently undergo crossover and to produce offspring for the next generation.

Crossover

In genetic algorithms, crossover is a recombination operator that exchanges genetic material between two selected parent solutions to produce , thereby simulating the inheritance of beneficial traits observed in and promoting diversity in the population. This process typically involves aligning the chromosomes of the parents and swapping segments at designated points, which allows the algorithm to combine advantageous building blocks from different individuals. The effectiveness of crossover relies on the prior selection of high-fitness parents, ensuring that the resulting progeny inherit promising genetic structures. Common types of crossover operators vary by representation and problem domain. For binary-encoded chromosomes, single-point crossover selects a single random locus and swaps the substrings beyond that point between parents, preserving large schema while introducing moderate disruption. Multi-point crossover extends this by using multiple random loci for swaps, increasing variability but risking greater disruption of building blocks. Uniform crossover, in contrast, exchanges each independently with a fixed probability, often 0.5, which provides fine-grained recombination suitable for exploring broad solution spaces. In permutation-based problems, such as the traveling salesman problem (TSP), order-preserving operators like partially mapped crossover (PMX) maintain the relative order of elements by mapping segments between two cut points and repairing the remaining positions to avoid duplicates. The crossover rate, typically set between 0.6 and 0.9, determines the probability of applying the operator to a pair of parents, balancing exploration and exploitation in the evolutionary search. For real-coded genetic algorithms addressing continuous optimization, crossover operators must handle numerical values without discretization artifacts, often leading to disruption if not designed carefully. Arithmetic crossover computes offspring as a convex combination of parents, given by
o=λp1+(1−λ)p2,o = \lambda p_1 + (1 - \lambda) p_2,
where λ\lambda is a fixed parameter (e.g., 0.5) or adaptively chosen, enabling interpolation within the search space. Blend crossover (BLX-α\alpha) addresses disruption by sampling offspring from an expanded interval around the parents: for each dimension, if p1<p2p_1 < p_2, the offspring is drawn uniformly from [p1−α(p2−p1),p2+α(p2−p1)][p_1 - \alpha (p_2 - p_1), p_2 + \alpha (p_2 - p_1)], with α\alpha typically 0.1 to promote neighborhood exploration. In binary representations, the schema theorem provides a theoretical foundation, stating that the expected number of copies of a schema HH after crossover satisfies
E[m(H,t+1)]≥m(H,t)(f(H)fˉ(t))(1−pcδ(H)l−1),E[m(H, t+1)] \geq m(H, t) \left( \frac{f(H)}{\bar{f}(t)} \right) (1 - p_c \frac{\delta(H)}{l-1}),
where m(H,t)m(H, t) is the number of instances at time tt, f(H)f(H) is the average fitness, fˉ(t)\bar{f}(t) is the population mean fitness, pcp_c is the crossover rate, δ(H)\delta(H) is the schema's defining length, and ll is the chromosome length; this implies that short, high-fitness schemata propagate with high probability under low-disruption crossover. An illustrative example occurs in function optimization, where crossover blends high-fitness parent vectors to produce offspring that inherit effective trait combinations, accelerating convergence toward global . In the TSP, PMX recombines route segments from two parent tours, mapping partial paths to preserve feasibility while exploring new permutations that leverage strong subsequences from each parent. These operators enhance the algorithm's ability to exploit selected genetic material, though their performance depends on tuning to the problem's structure.

Mutation

In genetic algorithms, mutation is a stochastic operator that introduces random alterations to one or more genes in an individual's , mimicking biological point to maintain population diversity and prevent premature convergence induced by crossover. This process ensures exploration of the search space beyond local optima by injecting novel genetic material, thereby balancing exploitation of promising solutions with broader sampling. Common mutation types vary by representation. For binary-encoded chromosomes, bit-flip mutation inverts a selected bit with probability pmp_m, typically set to 1/L1/L where LL is the chromosome length, to introduce small, targeted changes. In real-valued representations, Gaussian mutation adds noise drawn from a normal distribution N(0,σ)\mathcal{N}(0, \sigma) to each gene, where σ\sigma controls the perturbation magnitude and is often scaled by problem bounds or generation progress. For permutation-based encodings, such as in traveling salesman problems, swap mutation exchanges two randomly selected positions, while scramble mutation randomly reorders a contiguous subsequence to disrupt order without violating uniqueness. Mutation parameters emphasize subtlety to avoid degenerating into pure . Standard rates range from 0.001 to 0.01 per , ensuring rare but impactful changes that preserve selective from fitness evaluation. Adaptive strategies adjust pmp_m dynamically, increasing it during fitness stagnation to enhance or decreasing it in later generations for fine-tuning; for instance, rates may scale inversely with population diversity metrics. In artificial immune systems, hypermutation applies elevated rates—often exceeding 0.1—to antibody-like solutions with low affinity, accelerating in dynamic environments. Mathematically, the number of mutations per individual follows a with λ=Lpm\lambda = L p_m, modeling across genes. Within the , mutation disrupts schemata at a rate proportional to O(lpm)O(l p_m), where ll is the schema's defining length, quantifying how it erodes low-fitness patterns while allowing high-fitness ones to propagate. In neuroevolution, mutation slightly perturbs connection weights in evolving neural networks, such as adding small to explore architectural variations and improve performance on tasks like control problems.

Advanced Techniques

Operator Combinations

In genetic algorithms, core operators are integrated within a standard evolutionary cycle to drive improvement. The process typically begins with the random initialization of a of solutions, followed by fitness evaluation for each individual. Selection operators then identify high-performing individuals as parents, after which crossover recombines their genetic material and introduces random variations to . The new generation replaces the old , either fully in a generational model or partially in a steady-state approach, repeating until a termination criterion is met. This sequencing ensures that selection acts first to bias towards promising solutions, with crossover and mutation applied subsequently to survivors for variation—often with crossover applied to 60-90% of selected pairs and to all at a low per-locus probability. Parameter tuning plays a crucial role in balancing exploration and exploitation through operator rates and replacement strategies. High crossover rates (typically 0.6-1.0) emphasize exploitation by preserving and combining effective genetic building blocks from fit parents, while low mutation rates (0.001-0.01 per locus) promote exploration by injecting diversity to escape local optima. Generational replacement, which overhauls the entire population each iteration, fosters broad search but risks losing rare good solutions; in contrast, steady-state replacement updates only one or a few individuals per step, sustaining higher selection pressure and faster convergence in some domains. These choices must be adapted to problem characteristics, as overly aggressive exploitation can lead to premature convergence, while excessive exploration delays progress. The performance impacts of operator combinations are theoretically grounded in the schema theorem, which predicts the survival and growth of short, low-order (substrings representing partial solutions) with above-average fitness. Formally, m(H,t+1)≥m(H,t)f(H)fˉ(1−pcδ(H)l−1)(1−pm)lm(H,t+1) \geq m(H,t) \frac{f(H)}{\bar{f}} \left(1 - p_c \frac{\delta(H)}{l-1}\right) (1 - p_m)^l where m(H,t)m(H,t) denotes the number of instances of schema HH at generation tt, f(H)f(H) its average fitness, fˉ\bar{f} the population's mean fitness, pcp_c the crossover probability, δ(H)\delta(H) the defining length of HH, ll the chromosome length, and pmp_m the per position. This bound shows that moderate pcp_c and low pmp_m minimize disruption to promising schemata, enabling exponential growth of superior building blocks over time. Advanced strategies enhance operator integration for robustness. Elitist recombination explicitly carries over the best individuals unchanged to the next , guaranteeing non-decreasing fitness and accelerating convergence without additional computational overhead. Island models partition the population into semi-isolated subpopulations, each evolving via local operators, with periodic migration exchanging individuals to inject diversity and mitigate —effectively distributing operator effects across parallel search threads. In , for instance, the NSGA-II framework combines non-dominated sorting selection with simulated binary crossover (SBX), where SBX simulates binary recombination for real-valued parameters to maintain effective diversity while approximating Pareto-optimal fronts efficiently.

Variations and Extensions

Genetic operators have been adapted for specific problem domains to enhance their effectiveness in non-binary representations. For continuous search spaces, the simulated binary crossover (SBX) operator simulates binary crossover behavior while handling real-valued parameters, controlled by a distribution index η that influences the spread of offspring around parents. In permutation-based problems like the traveling salesman problem (TSP), edge recombination preserves adjacency information from parent tours by building child tours from shared edges, reducing the likelihood of producing invalid or low-quality solutions. Advanced extensions incorporate additional mechanisms to improve convergence and diversity. Memetic operators integrate local search techniques, such as hill-climbing, immediately after or crossover to refine solutions, blending global exploration with local exploitation. Co-evolutionary operators involve multiple subpopulations that evolve in parallel and interact through competitive or cooperative evaluations, allowing for the optimization of interdependent components in complex systems. Self-adaptive rates treat the mutation strength as an evolvable parameter within the itself, enabling the algorithm to dynamically adjust rates based on fitness feedback without external tuning. Hybrid approaches combine genetic operators with elements from other metaheuristics to address limitations in standard genetic algorithms. For instance, particle swarm optimization-inspired mutation updates individual positions using velocity and best-known solutions, as implemented in toolboxes like DEAP for enhanced exploration in continuous domains. Multi-operator genetic algorithms employ a of crossover and variants, switching among them based on detected stagnation—such as lack of fitness improvement over generations—to reinvigorate the search and escape local optima. Addressing challenges in constrained and large-scale optimization has driven further innovations. Repair operators post-mutation or crossover adjust infeasible solutions to satisfy constraints, such as by penalizing violations or regenerating valid structures, ensuring feasibility in problems like scheduling or engineering design. Parallel implementations distribute subpopulations across processors to handle large-scale problems, with migrations between islands promoting diversity; these gained prominence in the for scaling to thousands of variables in applications like . In , subtree crossover swaps subtrees between program trees to generate new expressions, preserving functional structure while introducing variation suitable for evolving code. Quantum-inspired operators leverage representations for superposition, allowing a single to encode multiple states probabilistically, with quantum gates simulating to explore solution spaces more efficiently in combinatorial tasks.

References

  1. https://direct.mit.edu/books/monograph/2574/Adaptation-in-Natural-and-Artificial-SystemsAn
  2. https://mitpress.mit.edu/9780262631853/an-introduction-to-genetic-algorithms/
  3. https://www.worldscientific.com/doi/abs/10.1142/S146902680000007X
  4. https://dl.acm.org/doi/abs/10.1145/3638530.3664086
  5. https://www.nature.com/articles/s41598-025-91245-z
  6. https://archive.org/details/theoryofselfrepr00vonn_0
  7. https://ieeexplore.ieee.org/document/5392654
  8. https://link.springer.com/book/9783540669535
  9. https://mitpress.mit.edu/9780262581110/adaptation-in-natural-and-artificial-systems/
  10. https://deepblue.lib.umich.edu/handle/2027.42/4507
  11. https://arxiv.org/pdf/1901.05737
  12. https://dl.acm.org/doi/book/10.5555/534133
  13. https://mitpress.mit.edu/9780262034011/genetic-programming/
  14. https://arxiv.org/abs/2010.03483
  15. https://ictactjournals.in/paper/IJSC_V6_I1_paper_4_pp_1083_1092.pdf
  16. https://www.researchgate.net/publication/288749263_CROSSOVER_OPERATORS_IN_GENETIC_ALGORITHMS_A_REVIEW
  17. https://ntrs.nasa.gov/api/citations/19900016189/downloads/19900016189.pdf
  18. https://2024.sci-hub.st/8144/147b4690afd2516b2a10edb8cb874ff8/eshelman1993.pdf
  19. https://www.researchgate.net/publication/318691362_Crossover_and_Mutation_Operators_of_Genetic_Algorithms
  20. https://link.springer.com/article/10.1007/s11042-020-10139-6
  21. https://www.egr.msu.edu/~kdeb/papers/k2012016.pdf
  22. https://www.tutorialspoint.com/genetic_algorithms/genetic_algorithms_mutation.htm
  23. https://www.eislab.gatech.edu/people/scholand/gapara.htm
  24. https://www.researchgate.net/publication/225642916_Adaptive_mutation_in_genetic_algorithms
  25. https://link.springer.com/chapter/10.1007/978-3-642-22371-6_14
  26. https://www.cs.jhu.edu/~ayuille/courses/Stat202C-Spring10/ga_tutorial.pdf
  27. https://archive.org/details/geneticalgorithm0000gold
  28. https://www.diva-portal.org/smash/get/diva2:832349/FULLTEXT01.pdf
  29. https://www.sciencedirect.com/science/article/pii/B9780080506845500094
  30. https://www.semanticscholar.org/paper/Genetic-Algorithms-in-Search-Optimization-and-Goldberg/2e62d1345b340d5fda3b092c460264b9543bc4b5
  31. https://link.springer.com/chapter/10.1007/978-3-642-28789-3_7
  32. https://sci2s.ugr.es/sites/default/files/files/Teaching/OtherPostGraduateCourses/Metaheuristicas/Deb_NSGAII.pdf
  33. https://www.semanticscholar.org/paper/Simulated-Binary-Crossover-for-Continuous-Search-Deb-Agrawal/b8ee6b68520ae0291075cb1408046a7dff9dd9ad
  34. https://www.complex-systems.com/abstracts/v09_i02_a02/
  35. https://www.researchgate.net/publication/2527549_The_Traveling_Salesman_and_Sequence_Scheduling_Quality_Solutions_Using_Genetic_Edge_Recombination
  36. https://www.researchgate.net/publication/267695201_Memetic_Algorithms
  37. https://link.springer.com/chapter/10.1007/3-540-58484-6_269
  38. https://www.researchgate.net/publication/225705383_Self-Adaptation_in_Evolutionary_Algorithms
  39. https://www.researchgate.net/publication/8337717_A_Hybrid_of_Genetic_Algorithm_and_Particle_Swarm_Optimization_for_Recurrent_Network_Design
  40. http://www.cs.ru.nl/~elenam/cspga.pdf
Add your contribution
Related Hubs
User Avatar
No comments yet.