Fighting noise with noise: DE with individuals shaking to tackle noisy problems

The idea of fighting noise with noise is introduced in this paper and it has been utilized to enhance Differential Evolution algorithm to solve noisy problems efficiently. The Monte-Carlo method is employed to investigate applicability of the proposed concept on a simple real-life problem. Based on the current concept, DE with population shaking (DEPS) and with individuals shaking (DEIS) are developed. Furthermore, the parent algorithm, DE, is experimentally compared with DEPS and DEIS on a benchmark test suite with nine well-known noisy functions. Detailed experimental verifications and corresponding analysis are presented for the 2D to 500D problems, for various noise levels and shaking rates.