Diversity-Guided Evolutionary Algorithms
PPSN VII Proceedings of the 7th International Conference on Parallel Problem Solving from Nature
Adaption to a Changing Environment by Means of the Thermodynamical Genetic Algorithm
PPSN IV Proceedings of the 4th International Conference on Parallel Problem Solving from Nature
Measurement of Population Diversity
Selected Papers from the 5th European Conference on Artificial Evolution
A generic approach to parameter control
EvoApplications'12 Proceedings of the 2012t European conference on Applications of Evolutionary Computation
Hi-index | 0.00 |
A number of diversity measures used in evolutionary computing suffer from 'mis-measuring' the diversity of populations. In this paper, we identify and demonstrate this problem using a carefully engineered test suite of six differently arranged populations. We also propose a new measure, called Population Diversity Index (PDI), that solves the problem. We show that sorting the test configurations by their PDI value we obtain a correct ranking (i.e., a natural one, conformant with the human-perceived order). PDI also allows for a comparison between populations of different sizes and genome-dimensions, and its relation to the uniform distribution makes the calculated diversity values easy to interpret.