On the Performance of Estimation of Distribution Algorithms Applied to Software Testing

Abstract

& One of the most important issues in software testing is the generation of the input cases used during the test. Due to the expensive cost of this task, its automation has become a key aspect. An alternative to obtain this is evolutionary testing. The aim of evolutionary testing is the creation of test data by means of combinatorial optimization search methods.A heuristic approach to the automatic generation of test cases is presented. The developed approach makes use of an emerging set of evolutionary algorithms called estimation of distribution algorithms. The analysis of the experimental results obtained presents this set of optimization techniques as a promising option for tackling this problem. More precisely, the performance of different estimation of distribution algorithms is evaluated and a comparison with the results of previous works is carried out.Testing is a crucial part of the software development process (Beizer 1990). It plays a main role in the search for the quality required as it constitutes the primary way used in practice to verify the correct behavior of the software produced. In fact, it is not unusual to dedicate at least 50% of the project resources to this phase (Boehm 1981).A major activity in testing is test design. Within this activity, a key task is the generation of the input cases to be applied to the program under test. This is a critical task, as the produced input must be adequate to the test type and its requirements. This, together with the fact that in most organizations this generation is performed manually, results in a high amount of resources dedicated to the test data generation step. Thus, the automatic

An evaluation of Differential Evolution in software test data generation

“…All the programmes we have selected but one are typical case studies used in some other work in the literature [6], [5], [18], [4], [19], [7]; namely, we had access to the source codes of triangle1, atof and remainder, which take integer parameters, and triangle2, which is the continuous parameters version of triangle1. In order to expand the comparison with realvalued domains, we added sncndn to the benchmark, which is a numerical calculus function extracted from [30].…”

“…The strategy to select the objective branch consists then of choosing the branch with a highest quality set of inputs. Moreover, for the optimisation step, this set is used to seed the initial phase of the search method [3], [4], [6]. This way, at each round, we try solve the optimisation problem with the most promising initial solutions for the search technique.…”

“…Step: A large amount of classical SB-STDG generators [3], [4], [5], [6] implicitly conform to a constraint-handling approach [7]. However, all of them handle the constraints in the order naturally imposed by this problem: the value of g i (x) is unknown unless g i−1 (x) = 0.…”

“…In order to have an idea of the performance of this technique compared to other population based evolutionary algorithms, we also experiment with the Breeder Genetic Algorithm [15], which is a well-known technique with a wide track record of successful real-world applications [16], [17]. Also, we face the DE based generator with other SBSTDG approaches from the literature [6], [5], [18], [4], [19], [7].…”

“…This problem is then addressed using search methods. Most of the SBSTDG approaches follow a dynamic strategy where the programme is executed and the information available at run-time is exploited to guide the search of test inputs [3], [4], [5], [6].…”