Impact of LD degree on the construction of Forests of Hierarchical Latent Class models. Five sequences showing variable LD degrees have been used to learn Forests of Hierarchical Latent Class Models. For display convention and node nomenclature, see Figure 6. We recall that in any FHLCM graph, edges are directed from top to bottom. a = 0.2, b = 2, card
= 20, t
= 0.95, t
(0.95), t = 0.6 (for CFHLC parameter description, see Section Algorithm).