We present a representation learning algorithm that learns a low-dimensional latent dynamical system from high-dimensional sequential raw data, e.g., video. The framework builds upon recent advances in amortized inference methods that use both an inference network and a refinement procedure to output samples from a variational distribution given an observation sequence, and takes advantage of the duality between control and inference to approximately solve the intractable inference problem using the path integral control approach. The learned dynamical model can be used to predict and plan the future states; we also present the efficient planning method that exploits the learned low-dimensional latent dynamics. Numerical experiments show that the proposed path-integral control based variational inference method leads to tighter lower bounds in statistical model learning of sequential data.

Paper: 5353 This paper discusses representation learning problem, and designs an algorithm to learn variational parameters of approximate inference. The paper has adopted a semi-amortized technique for learning variational parameters of dynamic system for temporal data. In this paper, dynamic system is generated by state space model. In order to express the state space model, just initial state and control input at every time steps needs to be inferred. Paper has utilized adaptive path-integral technique for variational parameter refinement of dynamic system to mitigate the amortization gap( induced by limitation of inference network parameters compared to optimal variational distribution).

We present a representation learning algorithm that learns a low-dimensional latent dynamical system from high-dimensional sequential raw data, e.g., video. The framework builds upon recent advances in amortized inference methods that use both an inference network and a refinement procedure to output samples from a variational distribution given an observation sequence, and takes advantage of the duality between control and inference to approximately solve the intractable inference problem using the path integral control approach. The learned dynamical model can be used to predict and plan the future states; we also present the efficient planning method that exploits the learned low-dimensional latent dynamics. Numerical experiments show that the proposed path-integral control based variational inference method leads to tighter lower bounds in statistical model learning of sequential data. Papers published at the Neural Information Processing Systems Conference.