Optimized Differential Evolution Algorithm for Software Testing

Differential evolution (DE) algorithms for software testing usually exhibited limited performance and stability owing to possible premature-convergence-related aging during evolution processes. This paper proposes a new framework comprising an antiaging mechanism, that is, a rebirth strategy with partial memory against aging, for the existing DE algorithm and a specialized fitness function. The results of application of the proposed framework to instantiate three DE algorithms with different mutation schemas indicate that it significantly improved their effectiveness, performance, and stability.