Emergent Communication through Negotiation

Cao, Kris, Lazaridou, Angeliki, Lanctot, Marc, Leibo, Joel Z, Tuyls, Karl and Clark, Stephen
(2018) Emergent Communication through Negotiation. 6th International Conference on Learning Representations, ICLR 2018 - Conference Track Proceedings, abs/18.

[img] Text
1804.03980v1.pdf - Submitted version

Download (542kB)


Multi-agent reinforcement learning offers a way to study how communication could emerge in communities of agents needing to solve specific problems. In this paper, we study the emergence of communication in the negotiation environment, a semi-cooperative model of agent interaction. We introduce two communication protocols -- one grounded in the semantics of the game, and one which is \textit{a priori} ungrounded and is a form of cheap talk. We show that self-interested agents can use the pre-grounded communication channel to negotiate fairly, but are unable to effectively use the ungrounded channel. However, prosocial agents do learn to use cheap talk to find an optimal negotiating strategy, suggesting that cooperation is necessary for language to emerge. We also study communication behaviour in a setting where one agent interacts with agents in a community with different levels of prosociality and show how agent identifiability can aid negotiation.

Item Type: Article
Additional Information: Published as a conference paper at ICLR 2018
Uncontrolled Keywords: cs.AI, cs.AI, cs.CL, cs.LG, cs.MA
Depositing User: Symplectic Admin
Date Deposited: 10 Dec 2018 15:21
Last Modified: 19 Jan 2023 01:09
Related URLs:
URI: https://livrepository.liverpool.ac.uk/id/eprint/3029649