EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

GEMA: Graph Embeddings for Multi‐Agent Coordination

Alessandro Amato, Raffaele Galliera, K. Brent Venable, Niranjan Suri

AI Magazine · 2026-09 · 47 (3)

DOI: 10.1002/aaai.70088 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AI Magazine, online erschienen am 24.08.2026. Verfasst von Alessandro Amato und 3 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 24 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Air Force Research Laboratory.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Many cooperative multi‐agent tasks are naturally defined by graph‐structured objectives, where agents must collectively reach, for example, a desired relational configuration or satisfy a set of constraints. These objectives often encode spatial arrangements, inter‐agent relations, or constraints that can be formalized as target graphs. However, current goal‐conditioned multi‐agent reinforcement learning (MARL) algorithms do not employ these symbolic and structured representations to direct their agents towards effective strategies. We propose Graph Embeddings for Multi‐Agent Coordination (GEMA), which couples any cooperative learner with a State‐Goal Graph Encoder (SGE). The SGE is pre‐trained in a contrastive manner to embed state graphs in a common metric space. At run time, each agent builds the state graph, queries the SGE, and computes a scalar distance to the goal embedding. The calculated distance can then be used as an intrinsic reward signal for the agents and incorporated into each agent's observation, providing feedback on the task progress. Experiments on challenging benchmarks for centralized training with decentralized execution (CTDE), including cooperative navigation, load balancing, and the StarCraft Multi‐Agent Challenge (v2), demonstrate that GEMA accelerates convergence and improves team returns, outperforming both standard and objective‐driven state‐of‐the‐art MARL baselines. Code repository: https://github.com/aamatodev/gema .

Förderung

  • Air Force Research Laboratory

Zitieren

APA

Amato, A., Galliera, R., Venable, K. B., & Suri, N. (2026). GEMA: Graph Embeddings for Multi‐Agent Coordination. AI Magazine, 47(3). https://doi.org/10.1002/aaai.70088

Vancouver

Amato A, Galliera R, Venable KB, Suri N. GEMA: Graph Embeddings for Multi‐Agent Coordination. AI Magazine. 2026;47(3). doi:10.1002/aaai.70088

BibTeX
@article{Amato2026,
  author = {Amato, Alessandro and Galliera, Raffaele and Venable, K. Brent and Suri, Niranjan},
  title = {GEMA: Graph Embeddings for Multi‐Agent Coordination},
  journal = {AI Magazine},
  year = {2026},
  volume = {47},
  number = {3},
  doi = {10.1002/aaai.70088}
}
Kostenlos starten →