## Niching Methods for Genetic Algorithms (1995)

Citations: | 208 - 1 self |

### BibTeX

@TECHREPORT{Mahfoud95nichingmethods,

author = {Samir W. Mahfoud},

title = {Niching Methods for Genetic Algorithms},

institution = {},

year = {1995}

}

### Years of Citing Articles

### OpenURL

### Abstract

Niching methods extend genetic algorithms to domains that require the location and maintenance of multiple solutions. Such domains include classification and machine learning, multimodal function optimization, multiobjective function optimization, and simulation of complex and adaptive systems. This study presents a comprehensive treatment of niching methods and the related topic of population diversity. Its purpose is to analyze existing niching methods and to design improved niching methods. To achieve this purpose, it first develops a general framework for the modelling of niching methods, and then applies this framework to construct models of individual niching methods, specifically crowding and sharing methods. Using a constructed model of crowding, this study determines why crowding methods over the last two decades have not made effective niching methods. A series of tests and design modifications results in the development of a highly effective form of crowding, called determin...

### Citations

3213 |
Genetic Programming: On the Programming of Computers by Means of Natural Selection
- Koza
- 1992
(Show Context)
Citation Context ... restricted to bit-string representations. Other possibilities include vectors of real numbers (L. Davis, 1991b; Eshelman & Schaffer, 1993; Goldberg, 1991a, 1991b), and high8 level computer programs (=-=Koza, 1992-=-). Although variable-length structures are appropriate for many problems, fixed-length structures are the norm. In this study, we restrict our attention to structures that are single strings of l bits... |

3066 |
Adaptation in natural and artificial systems’. The
- Holland
- 1975
(Show Context)
Citation Context ...hema theorem, and the building block hypothesis. Schemata While a GA on the surface processes strings, it implicitly processes schemata, which represent similarities between strings (Goldberg, 1989c; =-=Holland, 1992-=-). A GA can not, as a practical matter, visit every point in the search space. It can, however, sample a sufficient number of hyperplanes in highly fit regions of the search space. Each such hyperplan... |

2091 |
Genetic algorithms in search, optimization and machine learning
- Goldberg
- 1989
(Show Context)
Citation Context ... peak drops close to one, does the peak lose all representation in the population. Further studies experiment with both genotypic and phenotypic sharing, using similar test problems (Deb, 1989; Deb & =-=Goldberg, 1989-=-). Genotypic sharing employs Hamming distance as its distance measure, while Phenotypic sharing employs Euclidean distance in decoded parameter space. Phenotypic sharing gives slightly better results,... |

1326 | Handbook of Genetic Algorithms
- Davis
- 1991
(Show Context)
Citation Context ...mosome is typically a string of bits, so the term string is often used instead. GAs, however, are not restricted to bit-string representations. Other possibilities include vectors of real numbers (L. =-=Davis, 1991-=-b; Eshelman & Schaffer, 1993; Goldberg, 1991a, 1991b), and high8 level computer programs (Koza, 1992). Although variable-length structures are appropriate for many problems, fixed-length structures ar... |

929 | An Analysis of the Behavior of a Class of Genetic Adaptive Systems - DeJong - 1975 |

912 |
Adaptation in Natural and artificial systems. Ann Arbor
- Holland
- 1975
(Show Context)
Citation Context ...s to GA research. It discusses paths for future research, and draws overall conclusions from the research presented in this thesis. 7 Chapter 2 Genetic Algorithms Genetic algorithms (Goldberg, 1989c; =-=Holland, 1975-=-, 1992) are general purpose, parallel search procedures that are based upon genetic and evolutionary principles. A genetic algorithm works by repeatedly modifying a population of artificial structures... |

550 |
Genetic algorithms with sharing for multimodal function optimization
- Goldberg, Richardson
- 1987
(Show Context)
Citation Context ...one to characterize the functions that are most difficult for a GA to optimize (Bethke, 1980; Brindle, 1981; Das & Whitley, 1991; Davidor, 1991a; Deb & Goldberg, 1993, 1994; Forrest & Mitchell, 1993; =-=Goldberg, 1987-=-, 1989a, 1989b, 1991a, 1991b, 1992; Goldberg, Korb, & Deb, 1989; Grefenstette, 1993; Kargupta, Deb, & Goldberg, 1992; Kinnear, 1994; Liepins & Vose, 1990, 1991; Mason, 1991; Whitley, 1991). It is gene... |

499 | Genetic algorithms for multiobjective optimization: formulation, discussion - Fonseca, Fleming - 1993 |

478 | Very simple classification rules perform well on most commonly used datasets
- Holte
- 1993
(Show Context)
Citation Context ...and simple hillclimbers (L. Davis, 1991a). Likewise, many of the widely studied functions in the machine learning literature have been found relatively easy to learn with simple classification rules (=-=Holte, 1993-=-). With rare exception, studies of niching methods have used only a few of the simpler functions of this thesis. We have been able to differentiate the behavior of parallel niching genetic algorithms ... |

464 | Coevolving parasites improve simulated evolution as an optimization procedure - HILLIS - 1990 |

434 |
Optimization of control parameters for genetic algorithms
- Grefenstette
- 1986
(Show Context)
Citation Context ...ion pressure can be controlled through the use of fitness scaling.) A number of studies seek the optimal set of control parameters for a collection of test functions (Back, 1992, 1993; De Jong, 1975; =-=Grefenstette, 1986-=-; Schaffer, Caruana, Eshelman, & Das, 1989). Some of these studies are frequently misinterpreted as calls for a single set of control parameters to solve all problems. More careful readers have come t... |

423 | A comparative analysis of selection schemes used in genetic algorithms
- Goldberg, Deb
- 1991
(Show Context)
Citation Context ...the term string is often used instead. GAs, however, are not restricted to bit-string representations. Other possibilities include vectors of real numbers (L. Davis, 1991b; Eshelman & Schaffer, 1993; =-=Goldberg, 1991-=-a, 1991b), and high8 level computer programs (Koza, 1992). Although variable-length structures are appropriate for many problems, fixed-length structures are the norm. In this study, we restrict our a... |

361 |
Reducing Bias and Inefficiency in the Selection Algorithm
- Baker
(Show Context)
Citation Context ...solutions.) She proves that RWS has higher variance than five other fitnessproportionate selection schemes, and demonstrates RWS's inferior performance on several test functions. Other authors (J. E. =-=Baker, 1987-=-; Booker, 1982; De Jong, 1975) also obtain worse empirical results with RWS than with lower variance selection schemes. 28 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 GENERATIONS Figure 3.4: RWS runs on E5 , w... |

338 | Messy genetic algorithms: Motivation, analysis, and ¯rst results - Goldberg, Korb, et al. - 1989 |

316 | A niched Pareto genetic algorithm for multiobjective optimization - Horn, Nafpiolitis, et al. - 1994 |

287 |
The CHC Adaptive Search Algorithm: How to Have Safe Search When Engaging in Nontraditional Genetic Recombination. In: Foundations of Genetic Algorithms
- Eshelman
- 1991
(Show Context)
Citation Context ...ethods increase rather than decrease operator disruption. None of the three algorithms is capable of maintaining multiple, stable subpopulations within the same population. The CHC genetic algorithm (=-=Eshelman, 1991-=-; Eshelman & Schaffer, 1991), upon convergence, uses the best string in the population as a template to reinitialize the entire population. To form each element of the new population, the best string ... |

280 |
An investigation of niche and species formation in genetic function optimization
- Deb, Goldberg
- 1989
(Show Context)
Citation Context ... its contributions to GA research. It discusses paths for future research, and draws overall conclusions from the research presented in this thesis. 7 Chapter 2 Genetic Algorithms Genetic algorithms (=-=Goldberg, 1989-=-c; Holland, 1975, 1992) are general purpose, parallel search procedures that are based upon genetic and evolutionary principles. A genetic algorithm works by repeatedly modifying a population of artif... |

277 | Simulated Annealing and Boltzmann Machines A Stochastic Approach to Combinatorial Optimization and Neural Computing - Aarts, Korst - 1989 |

249 | Genetic algorithms, noise, and the sizing of populations - Goldberg, Deb, et al. - 1992 |

231 | Cognitive systems based on adaptive algorithms - Holland, Reitman - 1978 |

205 | Modeling genetic algorithms with Markov chains - Nix, Vose - 1992 |

199 | The Parallel Genetic Algorithm as Function Optimizer - MÜHLENBEIN, SCHOMISCH - 1991 |

162 |
Genetic algorithms and Walsh functions: Part II, deception and its analysis
- Goldberg
- 1989
(Show Context)
Citation Context ... its contributions to GA research. It discusses paths for future research, and draws overall conclusions from the research presented in this thesis. 7 Chapter 2 Genetic Algorithms Genetic algorithms (=-=Goldberg, 1989-=-c; Holland, 1975, 1992) are general purpose, parallel search procedures that are based upon genetic and evolutionary principles. A genetic algorithm works by repeatedly modifying a population of artif... |

159 |
Sizing Populations for Serial and Parallel Genetic Algorithms
- Goldberg
- 1989
(Show Context)
Citation Context ... its contributions to GA research. It discusses paths for future research, and draws overall conclusions from the research presented in this thesis. 7 Chapter 2 Genetic Algorithms Genetic algorithms (=-=Goldberg, 1989-=-c; Holland, 1975, 1992) are general purpose, parallel search procedures that are based upon genetic and evolutionary principles. A genetic algorithm works by repeatedly modifying a population of artif... |

155 | Fine-grained parallel genetic algorithms - MANDERICK, SPIESSENS - 1989 |

149 | Real-coded genetic algorithms and intervalschemata - Eshelman, Schaffer - 1993 |

148 |
Analyzing deception in trap functions
- Deb, Goldberg
- 1993
(Show Context)
Citation Context ...press). A line of GA research influential to this thesis is the methodology for analysis and design promoted by Goldberg and colleagues, based upon the Wright Brothers analogy mentioned in Chapter 1 (=-=Goldberg, 1993-=-a, 1993b, 1993c). In the remainder of the current chapter, we review this methodology, mention the possibility of fast-convergence proofs for GAs, and place this thesis into the broad spectrum of GA r... |

145 | A Sequential Niche Technique for Multimodal Function Optimization. EVol. Comput - Beasley, Bull, et al. - 1993 |

145 | How genetic algorithms really work: I. mutation and hillclimbing - Muhlenbein - 1992 |

133 |
Parallel genetic algorithms, population genetics and combinatorial optimization
- Mühlenbein
- 1989
(Show Context)
Citation Context ...nes convergence times for various distributed-GA architectures.) Predecessors, extensions, and further empirical tests of the PGA are described in a number of papers 42 (Gorges-Schleuter, 1989, 1991; =-=Muhlenbein, 1989-=-; Muhlenbein, Gorges-Schleuter, & Kramer, 1988; Muhlenbein, Schomisch, & Born, 1991a, 1991b). On some problems, the authors report a solution better than all previously published solutions. Gorges-Sch... |

131 | Selection in massively parallel genetic algorithms - Collins, Jefferson - 1991 |

131 | Relative BuildingBlock Fitness and the Building-Block Hypothesis
- Forrest, Mitchell
- 1992
(Show Context)
Citation Context ...Deb, & Clark, 1992; Mahfoud & Goldberg, 1995), a methodology is still needed for controlling convergence in the simple GA. 2.4 Research The most recent International Conference on Genetic Algorithms (=-=Forrest, 1993-=-) showcased the work of over 200 researchers, in both the theory and application of GAs. Theoretical research covered the modelling and analysis of GAs using Markov chains and other statistical method... |

129 |
Nonstationary Function Optimization using Genetic Algorithms with Dominance and Diploidy
- Goldberg, Smith
- 1987
(Show Context)
Citation Context ...one to characterize the functions that are most difficult for a GA to optimize (Bethke, 1980; Brindle, 1981; Das & Whitley, 1991; Davidor, 1991a; Deb & Goldberg, 1993, 1994; Forrest & Mitchell, 1993; =-=Goldberg, 1987-=-, 1989a, 1989b, 1991a, 1991b, 1992; Goldberg, Korb, & Deb, 1989; Grefenstette, 1993; Kargupta, Deb, & Goldberg, 1992; Kinnear, 1994; Liepins & Vose, 1990, 1991; Mason, 1991; Whitley, 1991). It is gene... |

126 | Multiobjective optimization using the niched Pareto Genetic Algorithm (Technical Report
- Horn, Nafpliotis
- 1993
(Show Context)
Citation Context ... relative proximity of all points in the search space. The perfect discrimination assumption goes by several names and has several consequences. The names are perfect discrimination, perfect sharing (=-=Horn, 1993-=-) (for fitness sharing), nonoverlapping niches (the partitioning of niches into equivalence classes already ensures they do not overlap), perfect comparison, and noiseless comparison. One consequence ... |

121 | Optimal mutation rates in genetic search - Back |

120 | Genetic algorithm for changing environments - Grefenstette - 1992 |

117 |
ASPARAGOS : An asynchronous parallel genetic optimization strategy
- GORGES-SCHLEUTER
- 1989
(Show Context)
Citation Context ...hwehm, 1992, empirically examines convergence times for various distributed-GA architectures.) Predecessors, extensions, and further empirical tests of the PGA are described in a number of papers 42 (=-=Gorges-Schleuter, 1989-=-, 1991; Muhlenbein, 1989; Muhlenbein, Gorges-Schleuter, & Kramer, 1988; Muhlenbein, Schomisch, & Born, 1991a, 1991b). On some problems, the authors report a solution better than all previously publish... |

116 | An Investigation into the Use of Hypermutation as an Adaptive Operator in Genetic Algorithms Having Continuous, Time-Dependent Nonstationary Environments - Cobb - 1990 |

116 | Massive multimodality, deception, and genetic algorithms. Parallel Problem Solving From Nature - Goldberg, Deb - 1992 |

114 | Finite Markov chain analysis of genetic algorithms - Goldberg, Segrest - 1987 |

110 | in Time and Space. The Parallel Genetic Algorithm- Foundtions of Genetic Algorithms
- Mühlenbein
- 1991
(Show Context)
Citation Context ...e few optima. For genotypic hillclimbing, also called bitclimbing, we set all neighborhood sizes to one bit. Both genotypic and phenotypic hillclimbers implement versions of next-ascent hillclimbing (=-=Muhlenbein, 1991-=-, 1992). Starting with a randomly chosen variable, they cycle through the variables, trying perturbations on each one. For the phenotypic hillclimber, a perturbation is either an upward or a downward ... |

109 |
Crowding and preselection revisited
- Mahfoud
- 1992
(Show Context)
Citation Context ...n, Mahfoud was able to determine why neither method was successful, and to develop design modifications that produced a successful crowding algorithm. The resulting algorithm, deterministic crowding (=-=Mahfoud, 1992-=-, 1994b), exhibited extensive niching capabilities when applied to both multimodal optimization problems and classification problems. Deterministic crowding has since been successfully applied to the ... |

107 | Rapid Accurate Optimization of Difficult Problems Using Fast Messy Genetic Algorithms - Goldberg, Deb, et al. - 1993 |

106 | A Variant of Evolution Strategies for Vector Optimization
- Kursawe
(Show Context)
Citation Context ... novel environments appears promising, as shown by Hillis (1990). (We review Hillis's work in the next section.) Preliminary application to dual-criteria optimization problems also appears promising (=-=Kursawe, 1991-=-). Given a stationary, single environment, however, most evidence points to the conclusion that a diploid GA's population will fully converge, although in time slower than the haploid GA's (Goldberg &... |

101 |
Epistasis Variance: A Viewpoint on GA-Hardness", in Foundations of Genetic Algorithms
- Davidor
- 1991
(Show Context)
Citation Context ...he difficulty of problems that the GA must solve. Much work has been done to characterize the functions that are most difficult for a GA to optimize (Bethke, 1980; Brindle, 1981; Das & Whitley, 1991; =-=Davidor, 1991-=-a; Deb & Goldberg, 1993, 1994; Forrest & Mitchell, 1993; Goldberg, 1987, 1989a, 1989b, 1991a, 1991b, 1992; Goldberg, Korb, & Deb, 1989; Grefenstette, 1993; Kargupta, Deb, & Goldberg, 1992; Kinnear, 19... |

100 | Punctuated equilibria : A parallel genetic algorithm - COHOON, HEGDE, et al. - 1987 |

98 | Genetic algorithms for tracking changing environments - Cobb, Grefenstette - 1993 |

96 |
The interaction of mutation rate, selection, and self adaptation within genetic algorithm
- Back
- 1992
(Show Context)
Citation Context ...ssproportionate selection, selection pressure can be controlled through the use of fitness scaling.) A number of studies seek the optimal set of control parameters for a collection of test functions (=-=Back, 1992-=-, 1993; De Jong, 1975; Grefenstette, 1986; Schaffer, Caruana, Eshelman, & Das, 1989). Some of these studies are frequently misinterpreted as calls for a single set of control parameters to solve all p... |

93 | Micro-Genetic Algorithms for Stationary and Non-Stationary Function Optimization, Intelligent Control and Adaptive Systems,Vol. 1196 - Krishnakumar - 1989 |

92 | Evolution Algorithms in Combinatorial Optimization - Muhlenbein, Gorges-Schleuter, et al. - 1988 |