Structured Inference for Recurrent Hidden Semi-markov Model.

Segmentation and labeling for high dimensional time series is an important yet challenging task in a number of applications, such as behavior understanding and medical diagnosis. Recent advances to model the nonlinear dynamics in such time series data has suggested involving recurrent neural networks into Hidden Markov Models. Despite the success, however, this involvement has caused the inference procedure much more complicated, often leading to intractable inference, especially for the discrete variables of segmentation and labeling. To achieve both flexibility and tractability in modeling nonlinear dynamics of discrete variables and to model both the long-term dependencies and the uncertainty of the segmentation labels, we inherits the Recurrent Hidden Semi-Markov Model and presents an effective bi-directional inference method. In detail, the proposed bi-directional inference network reparameterizes the categorical segmentation with the Gumbel-Softmax approximation and resorts to the Stochastic Gradient Variational Bayes. We evaluate the proposed model in a number of tasks, including speech modeling, automatic segmentation and labeling in behavior understanding, and sequential multi-objects recognition. Experimental results have demonstrated that our proposed model can achieve significant improvement over the state-of-the-art methods.