Abstract
The goal of this paper is to develop an effective approach allowing to capture accurately the intrinsic nature of data using an infinite shifted-scaled Dirichlet mixture model (InSSDMM). This article extends the finite statistical model to a more efficient multidimensional infinite mixture. The flexibility of the developed framework is demonstrated via some challenging medical applications that concern diabetic retinopathy detection in eye images and pneumonia identification in chest X-ray scans. The obtained results demonstrate the merits of such approach as compared to many other generative and discriminative models.