Automatic Design of a Hyper-Heuristic Framework With Gene Expression Programming for Combinatorial Optimization Problems

Abstract

Abstract-Hyper-heuristic approaches aim to automate heuristic design in order to solve multiple problems instead of designing tailor-made methodologies for individual problems. Hyper-heuristics accomplish this through a high level heuristic (heuristic selection mechanism and an acceptance criterion). This automates heuristic selection, deciding whether to accept or reject the returned solution. The fact that different problems or even instances, have different landscape structures and complexity, the design of efficient high level heuristics can have a dramatic impact on hyper-heuristic performance. In this work, instead of using human knowledge to design the high level heuristic, we propose a gene expression programming algorithm to automatically generate, during the instance solving process, the high level heuristic of the hyperheuristic framework. The generated heuristic takes information (such as the quality of the generated solution and the improvement made) from the current problem state as input and decides which low level heuristic should be selected and the acceptance or rejection of the resultant solution. The benefit of this framework is the ability to generate, for each instance, different high level heuristics during the problem solving process. Furthermore, in order to maintain solution diversity, we utilize a memory mechanism which contains a population of both high quality and diverse solutions that is updated during the problem solving process. The generality of the proposed hyper-heuristic is validated against six well known combinatorial optimization problem, with very different landscapes, provided by the HyFlex software. Empirical results comparing the proposed hyper-heuristic with state of the art hyper-heuristics, conclude that the proposed hyper-heuristic generalizes well across all domains and achieves competitive, if not superior, results for several instances on all domains.

An apprenticeship learning hyper-heuristic for vehicle routing in HyFlex

“…Finally, regarding the hyper-heuristic proposed in [23], it makes use of a long training time as well as an extensive tuning process. This is in contrast to ALHH in which the training time is very short and considered (for a fair comparison) while evaluating the performance of the algorithm.…”

“…This is in contrast to ALHH in which the training time is very short and considered (for a fair comparison) while evaluating the performance of the algorithm. Thus, comparing ALHH (and those approaches in Table II) to the method in [23] would not be fair and therefore this performance comparison has been discarded. …”

“…Drake et al [22] used grammatical evolution to generate components of a variable neighbourhood search approach for solving VRP. A recent work in [23] provided an evolutionary approach based on gene expression programming to generate selection hyper-heuristic components for cross domain search. The approach uses a long training time.…”

“…Also, it seems that the training is performed for each instance and thus the generated hyper-heuristic is not generalizable to unseen problem instances. Parameter tuning for each domain has been deemed necessary in [23] which lengthens the "training" time even further. Our method is different in that a considerably short training time is employed.…”

A Tabu Search hyper-heuristic strategy for t-way test suite generation

“…Recent developments have introduced hyper-heuristics that are able to automatically generate the LLHs, whereby the end user does not have to implement a set of LLHs for each problem domain. Moreover, the HLH is also able to evolve its own selection and acceptance criteria [21,22].…”