DeepMDP: Learning Continuous Latent Space Models for Representation Learning

Carles Gelada
Carles Gelada
Jacob Buckman
Jacob Buckman

International Conference on Machine Learning, pp. 2170-2179, 2019.

Cited by: 21|Bibtex|Views45
EI
Other Links: dblp.uni-trier.de|academic.microsoft.com|arxiv.org

Abstract:

Many reinforcement learning (RL) tasks provide the agent with high-dimensional observations that can be simplified into low-dimensional continuous states. To formalize this process, we introduce the concept of a DeepMDP, a parameterized latent space model that is trained via the minimization of two tractable losses: prediction of reward...More

Code:

Data:

Full Text
Your rating :
0

 

Tags
Comments