Genetic Algorithm for Finding Cluster Hierarchies

摘要: Hierarchical clustering algorithms have been studied extensively in the last years. However, existing approaches for hierarchical suffer from several drawbacks. The representation of results is often hard to interpret even large datasets. Many are not robust noise objects or overcome these limitation only by difficult parameter settings. As many heavily depend on their initialization, resulting get stuck a local optimum. In this paper, we propose novel geneticbased algorithm GACH (Genetic Algorithm finding Cluster Hierarchies) that solves those problems beneficial combination genetic algorithms, information theory and model-based clustering. capable find correct number model parameters using Minimum Description Length (MDL) principle does initialization use population-based stochastic search which ensures thorough exploration space. Moreover, outliers handled as they assigned appropriate inner nodes hierarchy root. An extensive evaluation synthetic well real data demonstrates superiority our over approaches.