AI Paper Atlas

Research map / Multi-agent systems and games / Cooperative multi-agent RL

Shield Decentralization for Safe Multi-Agent Reinforcement Learning

Daniel Melcer, Christopher Amato, Stavros Tripakis

NeurIPS 2022 · Multi-agent systems and games › Cooperative multi-agent RL · approach: Offline and networked MARL · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add decentralized shield instead of no shieldtabular Q-learning · Gridworld ISR fixed startviolations: 12.3 -> 0Improved
add decentralized shield instead of no shieldtabular Q-learning · Gridworld Pentagon fixed startviolations: 4.8 -> 0Improved
replace centralized shield with decentralized shieldtabular Q-learning · Gridworld ISR fixed startevaluation return: 90.4 -> 89.9No clear effect
replace centralized shield with decentralized shieldtabular Q-learning · Gridworld Pentagon random startevaluation return: 88.4 -> 80.3Hurt
replace centralized shield with decentralized shieldtabular Q-learning · Gridworld SUNY fixed startevaluation return: 83.3 -> 86.6Improved
replace centralized shield with decentralized shieldDQN · Particle-Momentum partial observability, random startevaluation return: 70.8 -> 80.5Improved
replace centralized shield with decentralized shieldDQN · Particle-Momentum full observability, fixed startevaluation return: 91.6 -> 91.6No clear effect
add decentralized shield instead of no shieldDQN · Particle-Momentum partial observability, random startevaluation return: 33.2 -> 80.5Improved
add decentralized shield instead of no shieldDQN · Particle-Momentum partial observability, random startviolations: 115.7 -> 0Improved
add decentralized shield instead of no shieldDQN · Particle-Momentum full observability, fixed startviolations: 3.0 -> 0Improved
decentralize the shield instead of centralizedshield synthesis · Gridworld collisionaverage safe joint actions: 23.30 -> 22.07No clear effect
decentralize the shield instead of centralizedshield synthesis · Particle-Momentumaverage safe actions: 21.67 -> 20.81No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy