From FMms to tms Start with several LCMs each based on a distinct subset of attributes Y3 Each partitions data from a certain perspective Different partitions are independent of each other ( x1)(X2)(X3)(X4)(X5(X6(x7 Link them up to form a tree model Get LTM Consider different perspectives in a single model X4 Multiple partitions of data that are correlated X1)(X2)(X3 Summary: An LTM can be viewed as a collections of FMMs, with their atent variables linked up to form a tree structure AAAl2014 Tutorial Nevin L Zhang HKUSTAAAI 2014 Tutorial Nevin L. Zhang HKUST 12 From FMMs to LTMs  Start with several LCMs,  Each based on a distinct subset of attributes  Each partitions data from a certain perspective.  Different partitions are independent of each other  Link them up to form a tree model  Get LTM  Consider different perspectives in a single model  Multiple partitions of data that are correlated. Summary: An LTM can be viewed as a collections of FMMs, with their latent variables linked up to form a tree structure