科研成果详情

In this paper, we introduce a nonparametric Bayesian approach for clustering based on both Dirichlet processes and generalized Dirichlet (GD) distribution. Thanks to the proposed approach, the obstacle of estimating the correct number of clusters is sidestepped by assuming an infinite number of components. The problems of overfitting and underfitting the data are also prevented due to the nature of the nonparametric Bayesian framework. The proposed model is learned through a variational method in which the whole inference process is analytically tractable with closed-form solutions. The effectiveness and merits of the proposed clustering approach are investigated on two challenging real applications namely anomaly intrusion detection and image spam filtering. © Springer-Verlag 2012.