An Improved Linear Population Size Reduction Based Parameters with Adaptive Learning Mechanism Differential Evolution (iLPALMDE) for Real-Parameter Single Objective Black Box Optimization

In this paper, we proposed an improved Linear population size reduction based Parameters with Adaptive Learning Mechanism Differential Evolution (iLPALMDE) for real-parameter single objective black box optimization. The LPALMDE algorithm is a state-of-the-art DE variant proposed recently, nevertheless, it still has some weakness, e.g. the update scheme of crossover rate is heavily dependent on the number of individuals in each group, which may fall into a bad adaptation of crossover rate when population size becomes small in the linear population size reduction scheme. Therefore, a new adaptation scheme of crossover rate was advanced in this paper to tackle the weakness. This novel improved algorithm is verified under the CEC2013 benchmarks, and experiment results shows that it was competitive with other state of the art DE variants.