Parameter Free Flower Algorithm Based Strategy for Pairwise Testing
Abdullah B. Nasser, Kamal Z. Zamli · 2018
Adopted to solve optimization problems, meta-heuristic algorithms aim to judiciously explore the search space in search of the good optimal solution. As such, the effectiveness of any particular meta-heuristic algorithm is heavily dependent on their control parameters, that is, to ensure balance exploration and exploitation. In the field of t-way testing, much work has been done to adopt meta-heuristic algorithms for generating interaction test suite (where t indicates the interaction strength). In this paper, we propose an Adaptive Flower Pollination Algorithm (AFPA) for pairwise testing. Unlike the original Flower Pollination Algorithm (FPA), our AFPA removes the static probability dependency inherent in FPA (i.e. for selection of local and global search operator). Specifically, we allow a dynamic and adaptive probability instead. The experimental results show that AFPA can produce the optimum results in many cases. AFPA also demonstrates its capacity to dynamically control global and local search based on the system configuration.