Why Advanced Population Initialization Techniques Perform Poorly in High Dimension?

Kazimipour, Borhan; Li, Xiaodong; Qin, A. K.

doi:10.1007/978-3-319-13563-2_41

Cited by 9 publications

(5 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A wide range of population initialization methods have been employed by evolutionary algorithms [95,96]. There are various conclusions, conflicting at times, on the effect of initialization methods on large-scale optimization [97][98][99]. Kazimipour et al [97] studied the effect of advanced initialization methods on large-scale optimization.…”

Section: Initialization Methodsmentioning

confidence: 99%

“…Kazimipour et al [99] used centered L 2 discrepancy to measure population uniformity as a function of population size and the dimensionality of the space. They reported that the loss of population uniformity (hence diversity) due to curse-of-dimensionality is the dominant factor in the performance degradation of optimization algorithms, regardless of the choice of the initialization method.…”

Section: Initialization Methodsmentioning

confidence: 99%

“…For example, it is well-known that the contrast in distance between randomly chosen points diminish as the dimensionality of the space increases [101,102], which has serious implications on various initialization/sampling techniques. Consequently, Kazimipour et al [99] recommended the use of advanced initialization methods only when the population size and the problem dimensionality are low.…”

Section: Initialization Methodsmentioning

confidence: 99%

See 2 more Smart Citations

A Review of Population-Based Metaheuristics for Large-Scale Black-Box Global Optimization—Part II

Omidvar

Yao

2022

IEEE Trans. Evol. Computat.

Self Cite

View full text Add to dashboard Cite

This paper is the second part of a two-part survey series on large-scale global optimization. The first part covered two major algorithmic approaches to large-scale optimization, namely decomposition methods and hybridization methods such as memetic algorithms and local search. In this part we focus on sampling and variation operators, approximation and surrogate modeling, initialization methods, and parallelization. We also cover a range of problem areas in relation to large-scale global optimization, such as multi-objective optimization, constraint handling, overlapping components, the component imbalance issue, and benchmarks, and applications. The paper also includes a discussion on pitfalls and challenges of current research and identifies several potential areas of future research.

show abstract

Section: Initialization Methodsmentioning

confidence: 99%

Section: Initialization Methodsmentioning

confidence: 99%

Section: Initialization Methodsmentioning

confidence: 99%

See 1 more Smart Citation

A Review of Population-Based Metaheuristics for Large-Scale Black-Box Global Optimization—Part II

Omidvar

Yao

2022

IEEE Trans. Evol. Computat.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Some initialisation methods use greedy algorithms that use a large population to select the best initial population subset (such as oppositionbased learning [OBL] and quasi-opposition-based learning [QBL] initialisers) [1,18]. Poor population uniformity at large dimensions is the main reason for the poor performance of advanced initialisers [19]. In contrast, some methods, such as chaotic numbers, low-discrepancy sequences methods, and QBLs, can improve the performance of meta-heuristic algorithms regardless of population size [20].…”

Section: Introductionmentioning

confidence: 99%

“…Convergence of the fitness function and the values of the constraints against the iterations in the cantilever beam design problem for the cuckoo search algorithm with the proposed initialisation method HASANZADEH AND KEYNIA -19 …”

mentioning

confidence: 99%

A new population initialisation method based on the Pareto 80/20 rule for meta‐heuristic optimisation algorithms

Hasanzadeh

Keynia

2021

IET Software

View full text Add to dashboard Cite

In this research, a new method for population initialisation in meta-heuristic algorithms based on the Pareto 80/20 rule is presented. The population in a meta-heuristic algorithm has two important tasks, including pushing the algorithm toward the real optima and preventing the algorithm from trapping in the local optima. Therefore, the starting point of a meta-heuristic algorithm can have a significant impact on the performance and output results of the algorithm. In this research, using the Pareto 80/20 rule, an innovative and new method for creating an initial population in meta-heuristic algorithms is presented. In this method, by using elitism, it is possible to increase the convergence of the algorithm toward the global optima, and by using the complete distribution of the population in the search spaces, the algorithm is prevented from trapping in the local optima. In this research, the proposed initialisation method was implemented in comparison with other initialisation methods using the cuckoo search algorithm. In addition, the efficiency and effectiveness of the proposed method in comparison with other wellknown initialisation methods using statistical tests and in solving a variety of benchmark functions including unimodal, multimodal, fixed dimensional multimodal, and composite functions as well as in solving well-known engineering problems was confirmed. | INTRODUCTIONThis is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

show abstract

Particle Swarms Reformulated Towards a Unified and Flexible Framework

Innocente

2021

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Why Advanced Population Initialization Techniques Perform Poorly in High Dimension?

Cited by 9 publications

References 16 publications

A Review of Population-Based Metaheuristics for Large-Scale Black-Box Global Optimization—Part II

A Review of Population-Based Metaheuristics for Large-Scale Black-Box Global Optimization—Part II

A new population initialisation method based on the Pareto 80/20 rule for meta‐heuristic optimisation algorithms

Particle Swarms Reformulated Towards a Unified and Flexible Framework

Contact Info

Product

Resources

About