Multidimensional continuous time Bayesian network classifiers

The multidimensional classification of multivariate time series deals with the assignment multiple classes to time-ordered data described by a set feature variables. Although this challenging task has received almost no attention in literature, it is present wide variety domains, such as medicine, finance or industry. complexity problem lies two nontrivial tasks, learning continuous and simultaneous class variables that may show dependencies between them. These can be addressed different strategies, but most them involve difficult preprocessing data, high space ignoring useful interclass dependencies. Additionally, been given development new classifiers based on probabilistic graphical models, even though transparent models facilitate further understanding domain. In paper, novel model proposed, which able classify discrete temporal sequence into while modeling their This extends Bayesian networks problem, are explicitly represent behavior evolve over time. Different methods for parameters structure presented, numerical experiments synthetic real-world encouraging results terms performance respect independent classifiers, current alternative approach under network paradigm.