Generic probabilistic prototype based classification of vectorial and proximity data

In supervised learning probabilistic models are attractive to define discriminative models in a rigid mathematical framework. More recently, prototype approaches, known for compact and efficient models, were defined in a probabilistic setting, but are limited to metric vectorial spaces. Here we propose a generalization of the discriminative probabilistic prototype learning algorithm for arbitrary proximity data, widely applicable to a multitude of data analysis tasks. We extend the algorithm to incorporate adaptive distance measures, kernels and non-metric proximities in a full probabilistic framework.