ABSTRACT
In this chapter we propose to define a description with distributions indexed by integers. By setting the range of indexes, we define the number of distributions used in the description. This way we define generic descriptions only depending on the range of the indexes, just as we fixed the number of iterations in a “for” loop. In pursuing this idea, we can redefine the notion of filter where each new evidence is incorporated into the result of a previous inference. If the index represents successive time intervals, we can then use these Markov to
networks within the Bayesian programming formalism.
