Beta DVBF: Learning State-Space Models for Control from High Dimensional Observations

11/02/2019 ∙ by Neha Das, et al. ∙ 8 ∙ share

Learning a model of dynamics from high-dimensional images can be a core ingredient for success in many applications across different domains, especially in sequential decision making. However, currently prevailing methods based on latent-variable models are limited to working with low resolution images only. In this work, we show that some of the issues with using high-dimensional observations arise from the discrepancy between the dimensionality of the latent and observable space, and propose solutions to overcome them.

READ FULL TEXT

Predictive Coding for Locally-Linear Control

High-dimensional observations and unknown dynamics are major challenges ...
03/02/2020 ∙ by Rui Shu, et al. ∙ 5 ∙ share

read it
Prediction, Consistency, Curvature: Representation Learning for Locally-Linear Control

Many real-world sequential decision-making problems can be formulated as...
09/04/2019 ∙ by Nir Levine, et al. ∙ 25 ∙ share

read it
DeepMDP: Learning Continuous Latent Space Models for Representation Learning

Many reinforcement learning (RL) tasks provide the agent with high-dimen...
06/06/2019 ∙ by Carles Gelada, et al. ∙ 4 ∙ share

read it
Gaussian-Dirichlet Random Fields for Inference over High Dimensional Categorical Observations

We propose a generative model for the spatio-temporal distribution of hi...
03/26/2020 ∙ by John E. San Soucie, et al. ∙ 0 ∙ share

read it
Heteroscedastic Uncertainty for Robust Generative Latent Dynamics

Learning or identifying dynamics from a sequence of high-dimensional obs...
08/18/2020 ∙ by Oliver Limoyo, et al. ∙ 0 ∙ share

read it
Learning the Linear Quadratic Regulator from Nonlinear Observations

We introduce a new problem setting for continuous control called the LQR...
10/08/2020 ∙ by Zakaria Mhammedi, et al. ∙ 4 ∙ share

read it
Controlled Gaussian Process Dynamical Models with Application to Robotic Cloth Manipulation

Over the last years, robotic cloth manipulation has gained relevance wit...
03/11/2021 ∙ by Fabio Amadio, et al. ∙ 0 ∙ share

read it

Beta DVBF: Learning State-Space Models for Control from High Dimensional Observations

Authors

Predictive Coding for Locally-Linear Control

Prediction, Consistency, Curvature: Representation Learning for Locally-Linear Control

DeepMDP: Learning Continuous Latent Space Models for Representation Learning

Gaussian-Dirichlet Random Fields for Inference over High Dimensional Categorical Observations

Heteroscedastic Uncertainty for Robust Generative Latent Dynamics

Learning the Linear Quadratic Regulator from Nonlinear Observations

Controlled Gaussian Process Dynamical Models with Application to Robotic Cloth Manipulation

This week in AI