Pith. sign in

REVIEW 3 cited by

Learning Multi-Agent Communication with Contrastive Learning

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2307.01403 v3 pith:6G6YEIN2 submitted 2023-07-03 cs.AI cs.LG

Learning Multi-Agent Communication with Contrastive Learning

classification cs.AI cs.LG
keywords learningcommunicationmessagescontrastiveeffectiveenvironmentinformationmethod
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Communication is a powerful tool for coordination in multi-agent RL. But inducing an effective, common language is a difficult challenge, particularly in the decentralized setting. In this work, we introduce an alternative perspective where communicative messages sent between agents are considered as different incomplete views of the environment state. By examining the relationship between messages sent and received, we propose to learn to communicate using contrastive learning to maximize the mutual information between messages of a given trajectory. In communication-essential environments, our method outperforms previous work in both performance and learning speed. Using qualitative metrics and representation probing, we show that our method induces more symmetric communication and captures global state information from the environment. Overall, we show the power of contrastive learning and the importance of leveraging messages as encodings for effective communication.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Cost-Effective Communication: An Auction-based Method for Language Agent Interaction

    cs.AI 2025-11 unverdicted novelty 6.0

    DALA treats inter-agent communication as a centralized auction where agents bid on message value density, yielding SOTA results on seven reasoning benchmarks with far fewer tokens than prior methods.

  2. Dynamic Generation of Multi-LLM Agents Communication Topologies with Graph Diffusion Models

    cs.CL 2025-10 unverdicted novelty 6.0

    GTD generates task-adaptive, sparse communication topologies for multi-LLM agents via guided iterative graph diffusion steered by a proxy model predicting accuracy, utility, and cost.

  3. SCALE-COMM: Shared, Contrastively-Aligned Latent Embeddings for MARL Communication

    cs.RO 2026-05 unverdicted novelty 5.0

    SCALE-COMM uses contrastive alignment on latent embeddings to decouple and stabilize communication learning from policy optimization in decentralized MARL, showing gains on benchmarks and a warehouse task.