Variational Temporal Abstraction

              Taesup Kim1,3, , Sungjin Ahn2 , Yoshua Bengio1
         1
           Mila, Université de Montréal, 2 Rutgers University, 3 Kakao Brain

                        Abstract
     We introduce a variational approach to learning and inference of temporally hierar-
     chical structure and representation for sequential data. We propose the Variational
     Temporal Abstraction (VTA), a hierarchical recurrent state space model that can
     infer the la ...