Automatically Designing More General Mutation Operators of Evolutionary Programming for Groups of Function Classes Using a Hyper-Heuristic

Abstract

In this study we use Genetic Programming (GP) as an offline hyper-heuristic to evolve a mutation operator for Evolutionary Programming. This is done using the Gaussian and uniform distributions as the terminal set, and arithmetic operators as the function set. The mutation operators are automatically designed for a specific function class. The contribution of this paper is to show that a GP can not only automatically design a mutation operator for Evolutionary Programming (EP) on functions generated from a specific function class, but also can design more general mutation operators on functions generated from groups of function classes. In addition, the automatically designed mutation operators also show good performance on new functions generated from a specific function class or a group of function classes.

A hyper-heuristic approach to automated generation of mutation operators for evolutionary programming

“…Hong et al [21] first demonstrated that GP could automatically construct random number generators which are typically used in EP. In a second paper, it was shown that ADMs could be trained on collections of functions classes, showing good performance across a broader range of functions [34], however a tradeoff between general training and specific performance was observed. This paper presents a study of the design of 23 ADMs, for 23 functions classes, and then tests each of the 23 ADMs on each of the function classes.…”

Hyper-heuristic approach: automatically designing adaptive mutation operators for evolutionary programming

“…Genetic programming (GP) [1] is a branch of evolutionary computation that can generate computer programs, and is widely applied in numerous fields [2][3][4][5][6][7][8][9][10]. Evolutionary programming (EP) is a black-box optimiser and mutation is the only operator in EP.…”

“…Researchers classify hyper-heuristics according to the feedback sources in the learning process: Online learning hyper-heuristics learn from a single instance of a problem; Offline learning hyper-heuristics learn from a set of training instances and generalise to unseen instances [18]. Both online [2][3][4][5] and offline hyper-heuristics [6][7][8][9][10] are applied to various research fields.…”

“…Shao et al [6] used multiobjective genetic programming as an offline learning hyper-heuristic to apply the feature learning for image classification. Hong et al [7,8] used GP as an offline learning hyper-heuristic to automatically design a mutation operator for EP and automatically design more general mutation operators for EP [9]. Ross et al [10] used an offline learning hyper-heuristic to represent a step towards a new method of using evolutionary algorithms that may solve some problems of acceptability for real-world use.…”

“…The contribution of this study is that this work realises the 'automatic' design of 'adaptive' mutation operators, it realises both 'automatic' and 'adaptive' at the same time has been revised and updated. Previous studies either automatically designed static/non-adaptive mutation operators [7][8][9], or human/manually designed adaptive mutation operators/mutation strategies for EP [13,16,17,[19][20][21][22]. To achieve this target, a set of adaptive factors was proposed, which collected and updated historical information during the evolutionary process.…”