Disentangled Sticky Hierarchical Dirichlet Process Hidden Markov Model

The Hierarchical Dirichlet Process Hidden Markov Model (HDP-HMM) has been used widely as a natural Bayesian nonparametric extension of the classical for learning from sequential and time-series data. A sticky HDP-HMM proposed to strengthen self-persistence probability in HDP-HMM. However, entangles strength prior transition together, limiting its expressiveness. Here, we propose more general model: disentangled (DS-HDP-HMM). We develop novel Gibbs sampling algorithms efficient inference this model. show that outperforms on both synthetic real data, apply new approach analyze neural data segment behavioral video