Timezone: »

 
Poster
Variational Temporal Abstraction
Taesup Kim · Sungjin Ahn · Yoshua Bengio

Wed Dec 11 05:00 PM -- 07:00 PM (PST) @ East Exhibition Hall B + C #151
in Deep Learning -- Generative Models »

We introduce a variational approach to learning and inference of temporally hierarchical structure and representation for sequential data. We propose the Variational Temporal Abstraction (VTA), a hierarchical recurrent state space model that can infer the latent temporal structure and thus perform the stochastic state transition hierarchically. We also propose to apply this model to implement the jumpy imagination ability in imagination-augmented agent-learning in order to improve the efficiency of the imagination. In experiments, we demonstrate that our proposed method can model 2D and 3D visual sequence datasets with interpretable temporal structure discovery and that its application to jumpy imagination enables more efficient agent-learning in a 3D navigation task.

Keywords: Recurrent Networks · Generative Models · Deep Learning · Deep Learning -> Predictive Models; Deep Learning
[ Paper]  [ Poster]  [ Slides] 

Author Information

Taesup Kim (Mila, Université de Montréal / Kakao Brain)
Sungjin Ahn (Rutgers University)
Yoshua Bengio (Mila - University of Montreal)

More from the Same Authors