The Wasserstein Believer: Learning Belief Updates for Partially Observable Environments through Reliable Latent Space Models

Onderzoeksoutput: Conference paper

7 Downloads (Pure)

Samenvatting

Partially Observable Markov Decision Processes (POMDPs) are used to model environments where the full state cannot be perceived by an agent. As such the agent needs to reason taking into account the past observations and actions. However, simply remembering the full history is generally intractable due to the exponential growth in the history space. Maintaining a probability distribution that models the belief over what the true state is can be used as a sufficient statistic of the history, but its computation requires access to the model of the environment and is often intractable. While SOTA algorithms use Recurrent Neural Networks to compress the observation-action history aiming to learn a sufficient statistic, they lack guarantees of success and can lead to sub-optimal policies. To overcome this, we propose the Wasserstein Belief Updater, an RL algorithm that learns a latent model of the POMDP and an approximation of the belief update. Our approach comes with theoretical guarantees on the quality of our approximation ensuring that our outputted beliefs allow for learning the optimal value function.
Originele taal-2English
TitelThe Twelfth International Conference on Learning Representations
SubtitelICLR 2024
UitgeverijOpenReview.net
Aantal pagina's9
StatusPublished - 7 mei 2024
EvenementThe Twelfth International Conference on Learning Representations - Messe Wien Exhibition and Congress Center, Vienna, Austria
Duur: 7 mei 202411 mei 2024
Congresnummer: 12
https://iclr.cc

Conference

ConferenceThe Twelfth International Conference on Learning Representations
Verkorte titelICLR 2024
Land/RegioAustria
StadVienna
Periode7/05/2411/05/24
Internet adres

Keywords

  • cs.LG
  • cs.AI

Vingerafdruk

Duik in de onderzoeksthema's van 'The Wasserstein Believer: Learning Belief Updates for Partially Observable Environments through Reliable Latent Space Models'. Samen vormen ze een unieke vingerafdruk.

Citeer dit