Emergent Communication: Generalization and Overfitting in Lewis Games - Inria - Institut national de recherche en sciences et technologies du numérique
Conference Papers Year : 2022

Emergent Communication: Generalization and Overfitting in Lewis Games

Abstract

Lewis signaling games are a class of simple communication games for simulating the emergence of language. In these games, two agents must agree on a communication protocol in order to solve a cooperative task. Previous work has shown that agents trained to play this game with reinforcement learning tend to develop languages that display undesirable properties from a linguistic point of view (lack of generalization, lack of compositionality, etc). In this paper, we aim to provide better understanding of this phenomenon by analytically studying the learning problem in Lewis games. As a core contribution, we demonstrate that the standard objective in Lewis games can be decomposed in two components: a co-adaptation loss and an information loss. This decomposition enables us to surface two potential sources of overfitting, which we show may undermine the emergence of a structured communication protocol. In particular, when we control for overfitting on the co-adaptation loss, we recover desired properties in the emergent languages: they are more compositional and generalize better.
Fichier principal
Vignette du fichier
2209.15342.pdf (1.54 Mo) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-03831908 , version 1 (27-10-2022)

Identifiers

Cite

Mathieu Rita, Corentin Tallec, Paul Michel, Jean-Bastien Grill, Olivier Pietquin, et al.. Emergent Communication: Generalization and Overfitting in Lewis Games. NeurIPS 2022 - Thirty-sixth Conference on Neural Information Processing Systems, Nov 2022, Nouvelle-Orléans, United States. ⟨hal-03831908⟩
38 View
76 Download

Altmetric

Share

More