HIGhER: Improving instruction following with Hindsight Generation for Experience Replay - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2020

HIGhER: Improving instruction following with Hindsight Generation for Experience Replay

Mathieu Seurin
  • Function : Author
  • PersonId : 1039295
Florian Strub
Olivier Pietquin

Abstract

Language creates a compact representation of the world and allows the description of unlimited situations and objectives through compositionality. While these characterizations may foster instructing, conditioning or structuring interactive agent behavior, it remains an open-problem to correctly relate language understanding and reinforcement learning in even simple instruction following scenarios. This joint learning problem is alleviated through expert demonstrations, auxiliary losses, or neural inductive biases. In this paper, we propose an orthogonal approach called Hindsight Generation for Experience Replay (HIGhER) that extends the Hindsight Experience Replay approach to the language-conditioned policy setting. Whenever the agent does not fulfill its instruction, HIGhER learns to output a new directive that matches the agent trajectory, and it relabels the episode with a positive reward. To do so, HIGhER learns to map a state into an instruction by using past successful trajectories, which removes the need to have external expert interventions to relabel episodes as in vanilla HER. We show the efficiency of our approach in the BabyAI environment, and demonstrate how it complements other instruction following methods.
Fichier principal
Vignette du fichier
HIGhER___ADPRL.pdf (1.02 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03123981 , version 1 (28-01-2021)

Identifiers

  • HAL Id : hal-03123981 , version 1

Cite

Geoffrey Cideron, Mathieu Seurin, Florian Strub, Olivier Pietquin. HIGhER: Improving instruction following with Hindsight Generation for Experience Replay. ADPRL 2020 - IEEE SSCI Conference on Adaptive Dynamic Programming and Reinforcement Learning, Dec 2020, Camberra / Virtual, Australia. ⟨hal-03123981⟩
96 View
192 Download

Share

Gmail Facebook X LinkedIn More