Résumé : We address the issue of classifying complex data. We focus on three main sources of complexity, namely the high dimensionality of the observed data, the dependencies between these observations and the general nature of the noise model underlying their distribution. We investigate the recent \textit{Triplet Markov Fields} and propose new models in this class that can model such data and handle the additional inclusion of a learning step in a consistent way. One advantage of our models is that their estimation can be carried out using state-of-the-art Bayesian clustering techniques. As generative models, they can be seen as an alternative, in the supervised case, to discriminative Conditional Random Fields. Identifiability issues and possible phase transition phenomena underlying the models in the non supervised case, are discussed while the models performance is illustrated on real data exhibiting the mentioned various sources of complexity.