Dynamic Bayesian Network Action Set

cliques Data Table

The cliques data table is optional and specifies the cliques in the clique tree that are used to update beliefs from one time period to the next. If these cliques are not specified, then the action processes the Bayesian network to obtain the clique tree. The outCliques parameter creates an output data table that defines cliques. Subsequent runs of the filtering algorithm in later time periods use this data table as input to obtain cliques, so the action does not need to process the Bayesian network in order to generate the cliques in each run of the action. The cliques data table has the following columns:

id: specifies the clique or factor index that is used to identify a clique or factor.
type: specifies whether the index is for a clique or for one of several types of factors, such as incoming belief, conditional distribution, message, and so on.
varname: specifies the variable in the clique or factor.
stage: specifies the time period for the varname variable: 1 for the current time period, and 2 for the previous time period.
toid: specifies the clique index that this factor belongs to, if the type column indicates that it is a factor. If the type column indicates that it is a factor of the message type, then toid specifies the clique that receives this message.

Last updated: August 04, 2026