# Message-Dropout: An Efficient Training Method for Multi-Agent Deep Reinforcement Learning

arXiv: https://arxiv.org/abs/1902.06527 · HTML: https://mlsanity.com/paper/1902.06527/

## Key facts

- Cited by 17 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Feb 2019.

## Built on by (most cited first, 17 of 17)

- [A Survey of Multi-Agent Deep Reinforcement Learning with Communication](https://mlsanity.com/paper/2203.08975/) (2022)
- [Deep Multiagent Reinforcement Learning: Challenges and Directions](https://mlsanity.com/paper/2106.15691/) (2021)
- [Multi-Agent Reinforcement Learning for Autonomous Driving: A Survey](https://mlsanity.com/paper/2408.09675/) (2024)
- [Parameter Sharing with Network Pruning for Scalable Multi-Agent Deep Reinforcement Learning](https://mlsanity.com/paper/2303.00912/) (2023)
- [HAMMER: Multi-Level Coordination of Reinforcement Learning Agents via Learned Messaging](https://mlsanity.com/paper/2102.00824/) (2021)
- [A Variational Approach to Mutual Information-Based Coordination for Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2303.00451/) (2023)
- [The Power of Communication in a Distributed Multi-Agent System](https://mlsanity.com/paper/2111.15611/) (2021)
- [Centralized Training with Hybrid Execution in Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2210.06274/) (2022)
- [Fully Independent Communication in Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2401.15059/) (2024)
- [Multi-Agent Reinforcement Learning with Communication-Constrained Priors](https://mlsanity.com/paper/2512.03528/) (2025)
- [Improving the Robustness and Generalization of Deep Neural Network with Confidence Threshold Reduction](https://mlsanity.com/paper/2206.00913/) (2022)
- [Curiosity-driven Exploration in Sparse-reward Multi-agent Reinforcement Learning](https://mlsanity.com/paper/2302.10825/) (2023)
- [FlexiDrop: Theoretical Insights and Practical Advances in Random Dropout Method on GNNs](https://mlsanity.com/paper/2405.20012/) (2024)
- [Multi-Robot Pursuit in Parameterized Formation via Imitation Learning](https://mlsanity.com/paper/2410.23586/) (2024)
- [Q-MARL: A quantum-inspired algorithm using neural message passing for large-scale multi-agent reinforcement learning](https://mlsanity.com/paper/2503.07397/) (2025)
- [Communication Gain and Delay Cost Under Cross-Timestep Delays in Cooperative Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2604.03785/) (2026)
- [Value-Aware Prediction for Robust Multi-Agent Coordination Under Communication Loss](https://mlsanity.com/paper/2607.17914/) (2026)

## Abstract (arXiv)

In this paper, we propose a new learning technique named message-dropout to improve the performance for multi-agent deep reinforcement learning under two application scenarios: 1) classical multi-agent reinforcement learning with direct message communication among agents and 2) centralized training with decentralized execution. In the first application scenario of multi-agent systems in which direct message communication among agents is allowed, the message-dropout technique drops out the received messages from other agents in a block-wise manner with a certain probability in the training phase and compensates for this effect by multiplying the weights of the dropped-out block units with a correction probability. The applied message-dropout technique effectively handles the increased input dimension in multi-agent reinforcement learning with communication and makes learning robust against communication errors in the execution phase. In the second application scenario of centralized training with decentralized execution, we particularly consider the application of the proposed message-dropout to Multi-Agent Deep Deterministic Policy Gradient (MADDPG), which uses a centralized critic to train a decentralized actor for each agent. We evaluate the proposed message-dropout technique for several games, and numerical results show that the proposed message-dropout technique with proper dropout rate improves the reinforcement learning performance significantly in terms of the training speed and the steady-state performance in the execution phase.