Browsing by Author "Kononova, A.V."

Now showing 1 - 3 of 3
  • No Thumbnail Available
    ItemMetadata only
    Differential Evolution with Scale Factor Local Search for Large Scale Problems
    (SpringerLink, 2010) Caponio, A.; Kononova, A.V.; Neri, Ferrante
    This chapter proposes the integration of fitness diversity adaptation techniques within the parameter setting of Differential Evolution (DE). The scale factor and crossover rate are encoded within each genotype and self-adaptively updated during the evolution by means of a probabilistic criterion which takes into account the diversity properties of the entire population. The population size is also adaptively controlled by means of a novel technique based on a measurement of the fitness diversity. An extensive experimental setup has been implemented by including multivariate problems and hard to solve fitness landscapes. A comparison of the performance has been conducted by considering a standard DE as well as modern DE based algorithms, recently proposed in literature. Numerical results available show that the proposed approach seems to be very promising for some fitness landscapes and still competitive with modern algorithms in other cases. In most cases analyzed the proposed self-adaptation is beneficial in terms of algorithmic performance and can be considered a useful tool for enhancing the performance of a DE scheme.
  • Thumbnail Image
    ItemOpen Access
    Structural Bias in Differential Evolution: a preliminary study
    (2018-09-18) Caraffini, Fabio; Kononova, A.V.
    This paper extends the study of structural bias in popular metaheuristic global optimisation methods. Previously, it has been shown that both Genetic Algorithms and Particle Swarm Optimisation suffer from such bias. This means that difficulties already posed for a structurally biased algorithm by the fitness landscape itself are further unnecessarily exacerbated by the unexpected oversampling of some regions of the search space and avoidance of the others, to potential great detriment of the overall optimisation performance. Such bias is inherent in the core design of the algorithm. After careful examination, the authors conclude that some variants of Differential Evolution are not free of the structural bias. However, investigation suggests that the mechanisms of the formation of structural bias in Differential Evolution is different and can be balanced through a more careful design.
  • Thumbnail Image
    ItemEmbargo
    Structural bias in population-based algorithms
    (Elsevier, 2014-12-04) Kononova, A.V.; Corne, David W.; De Wilde, Philippe; Shneer, Vsevolod; Caraffini, Fabio
    Abstract Challenging optimisation problems are abundant in all areas of science and industry. Since the 1950s, scientists have responded to this by developing ever-diversifying families of ‘black box’ optimisation algorithms. The latter are designed to be able to address any optimisation problem, requiring only that the quality of any candidate solution can be calculated via a ‘fitness function’ specific to the problem. For such algorithms to be successful, at least three properties are required: (i) an effective informed sampling strategy, that guides the generation of new candidates on the basis of the fitnesses and locations of previously visited candidates; (ii) mechanisms to ensure efficiency, so that (for example) the same candidates are not repeatedly visited; and (iii) the absence of structural bias, which, if present, would predispose the algorithm towards limiting its search to specific regions of the solution space. The first two of these properties have been extensively investigated, however the third is little understood and rarely explored. In this article we provide theoretical and empirical analyses that contribute to the understanding of structural bias. In particular, we state and prove a theorem concerning the dynamics of population variance in the case of real-valued search spaces and a ‘flat’ fitness landscape. This reveals how structural bias can arise and manifest as non-uniform clustering of the population over time. Critically, theory predicts that structural bias is exacerbated with (independently) increasing population size, and increasing problem difficulty. These predictions, supported by our empirical analyses, reveal two previously unrecognised aspects of structural bias that would seem vital for algorithm designers and practitioners. Respectively, (i) increasing the population size, though ostensibly promoting diversity, will magnify any inherent structural bias, and (ii) the effects of structural bias are more apparent when faced with (many classes of) ‘difficult’ problems. Our theoretical result also contributes to the ‘exploitation/exploration’ conundrum in optimisation algorithm design, by suggesting that two commonly used approaches to enhancing exploration – increasing the population size, and increasing the disruptiveness of search operators – have quite distinct implications in terms of structural bias.