Question about WeakLimit HDP-HSMM: #119
Description
Hi everyone,
thank you for this repository. It is very helpful.
However, I am having trouble understanding the WeakLimitHDPHSMM function. As far as I understand it is an adaptation of Fox's blocked gibbs sampling algorithm (Fox et al 2010 see sticky_hdp_hmm for sticky HDP-HMM for HDP-HSMM using geometrically distributed auxiliary variables, but i do not understand the full algorithm. Is there some kind of pseudocode for this? For example, does it also use the chinese restaurant franchise?
That would be very helpful to understand the whole algorithm.
Thank you very much.