A General Framework for Statistical Performance Comparison of Evolutionary Computation Algorithms

D. Shilane (USA), J. Martikainen (Finland), S. Dudoit (USA), and S.J. Ovaska (Finland)


Evolutionary computation, statistics, performance comparison, two-fold sampling, bootstrap, multiple hypothesis testing.


This paper proposes a statistical methodology for comparing the performance of evolutionary computation algorithms. A two-fold sampling scheme for collecting performance data is introduced, and this data is assessed using a multiple hypothesis testing framework relying on a bootstrap resampling procedure. The proposed method offers a convenient, flexible, and reliable approach to comparing algorithms in a wide variety of applications.

Important Links:

Go Back