A multi-objective and evolutionary hyper-heuristic applied to the Integration and Test Order Problem

Abstract

The field of Search-Based Software Engineering (SBSE) has widely utilized Multi-Objective Evolutionary Algorithms (MOEAs) to solve complex software engineering problems. However, the use of such algorithms can be a hard task for the software engineer, mainly due to the significant range of parameter and algorithm choices. To help in this task, the use of Hyper-heuristics is recommended. Hyper-heuristics can select or generate low-level heuristics while optimization algorithms are executed, and thus can be generically applied. Despite their benefits, we find only a few works using hyper-heuristics in the SBSE field. Considering this fact, we describe HITO, a Hyper-heuristic for the Integration and Test Order problem, to adaptively select search operators while MOEAs are executed using one of the selection methods: Choice Function and Multi-Armed Bandit. The experimental results show that HITO can outperform the traditional MOEAs NSGA-II and MOEA/DD. HITO is also a generic algorithm, since the user does not need to select crossover and mutation operators, nor adjust their parameters.

A Hyper-heuristic for Multi-objective Integration and Test Ordering in Google Guava

“…For example, in an object-oriented system units are classes, in component-based programming units are components, in aspectoriented programming units are aspects, and in product line oriented systems units may be considered product features [2]. These characteristics make this problem suitable for the application of meta-and hyper-heuristics, since these kinds of algorithms are capable of optimizing several objectives at once [3], and are capable of being easily applied to different contexts without needing to have their implementation adapted for this end [7].…”

“…A further performance comparison of using HITO within an SPEA2 framework was given in a later paper [6]. Guizzo et al [7] provided another extension, formulating the problem as a many-objective problem with four objectives, comparing to a number of state-of-the-art MOEAs.…”

“…Google Guava [1] is a large open-source project, containing Google versions of a number of standard general purpose libraries for Java. In this paper, we use all three versions of HITO presented by Guizzo et al [7], to search for an optimal ordering of units for integration testing of Guava. A performance comparison to the original NSGA-II method presented by Assunção et al [2] is given.…”

A Novel Hyper-Heuristic for the Biobjective Regional Low-Carbon Location-Routing Problem with Multiple Constraints

“…They integrated of the HH component, credit assignment, and selection method (i.e., random and upper confidence bound) into MOEAs, that is, NSGA-II, SPEA2, indicator-based evolutionary algorithm (IBEA), and MOEA based on decomposition (MOEA/D). Guizzo et al [56] designed their MOHH to tackle integration and test ordering in Google Guava by inserting it MOHH into the MOEAs framework to evaluate the operators' performance through CF and MAB. Yao et al [57] proposed a MOHH framework for walking route planning in a smart city and a reinforcement learning mechanism was established to select the LLH.…”

Tuning Multi-Objective Evolutionary Algorithms on Different Sized Problem Sets

“…The problem sets contained different numbers of previously mentioned systems for the ITO problem. For each system, the stopping criterion was set to 60,000 fitness evaluations [9,29] The number of repetitions in MOCRS-Tuning was set to 8. The control parameters which underwent the tuning process were population size, crossover probability, and mutation probability.…”