Epigenetically Inspired Genetic Algorithm for Clustering Data

The paper presents a new genetic algorithm (GA) for clustering data (Epi_GGA). In the proposed algorithm, apart from standard operations for genetic algorithms, there are two additional operations mimicking epigenetic processes. In real world epigenetic processes they are cause of faster and better adapt individuals to the changing environment. So, the intention of the use of epigenetic processes in genetic algorithm is to speed up working time by reducing the number of iterations needed to obtain expected result. In the presented algorithm, the AIC criterion, the Hausdorff distance and the density of data were used to construct the fitness function. The proposed algorithm achieves very good results in solving the problem of clustering data, what was shown in described experiments.