Learning Distributed Cooperative Policies for Security Games via Deep Reinforcement Learning

Sheikh, Hassam Ullah; Razghandi, Mina; Bölöni, Ladislau

doi:10.1109/compsac.2019.00075

Cited by 4 publications

(6 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We perform our experiments using the defensive escort problem on four VIP protection environments [16], [3]. This is a medium-size collaborative problem where a defensive escort team of agents is learning to maintain an optimal formation For each agent i, select action a t i = π θi (o t i )…”

Section: A Environmentsmentioning

confidence: 99%

“…Cooperative multi-agent problems are prevalent in realworld settings such as strategic conflict resolution [1], coordination between autonomous vehicles [2] and collaboration of agents in defensive escort teams [3]. Such problems can be modelled as dual-interest: each agent is simultaneously working towards maximizing its own payoff (local reward) as well as the collective success of the team (global reward).…”

Section: Introductionmentioning

confidence: 99%

“…This was not previously feasible in the multi-agent RL setting. • We evaluate our proposed solution on the defensive escort team problem [3], [16] (see Figure 1) and show that it achieves a significantly better and more stable performance than the direct adaptation of the MADDPG algorithm.…”

Section: Introductionmentioning

confidence: 99%

“…Furthermore, previous work using these environments provide strong baselines against which the proposed algorithms can be compared. Aligning our experimental setup with [3], [16], we chose 4 bodyguards and 10 bystanders.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Multi-Agent Reinforcement Learning for Problems with Combined Individual and Team Reward

Sheikh¹,

Bölöni²

2020

Preprint

Self Cite

View full text Add to dashboard Cite

Many cooperative multi-agent problems require agents to learn individual tasks while contributing to the collective success of the group. This is a challenging task for current state-of-the-art multi-agent reinforcement algorithms that are designed to either maximize the global reward of the team or the individual local rewards. The problem is exacerbated when either of the rewards is sparse leading to unstable learning. To address this problem, we present Decomposed Multi-Agent Deep Deterministic Policy Gradient (DE-MADDPG): a novel cooperative multi-agent reinforcement learning framework that simultaneously learns to maximize the global and local rewards. We evaluate our solution on the challenging defensive escort team problem and show that our solution achieves a significantly better and more stable performance than the direct adaptation of the MADDPG algorithm.

show abstract

Section: A Environmentsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Multi-Agent Reinforcement Learning for Problems with Combined Individual and Team Reward

Sheikh¹,

Bölöni²

2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…It also has applications in home care, military combat and mobile sensor networks. For example cooperative bodyguards can be of good use [26] [27]. Pursuit evasion is widely researched in differentials games as well [15] [16].…”

Section: Related Workmentioning

confidence: 99%

A novel approach for multi-agent cooperative pursuit to capture grouped evaders

et al. 2018

View full text Add to dashboard Cite

Cite this article: Qadir, M.Z., Piao, S., Jiang, H. et al. A novel approach for multi-agent cooperative pursuit to capture grouped evaders. J Supercomput 76, 3416-3426 (2020). https://doi. AbstractAn approach of mobile multi-agent pursuit, based on application of Self-Organizing Feature Map (SOFM) and along with that reinforcement learning based on Agent Group Role Membership Function (AGRMF) model is proposed. This method promotes dynamic organization of the pursuers' groups and also makes pursuers' group evader according to their desire based on SOFM and AGRMF techniques. This helps to overcome the shortcomings of the pursuers that they cannot fully reorganize when the goal is too independent in process of AGRMF models operation. Besides, we also discuss a new reward function. After the formation of the group, reinforcement learning is applied to get the optimal solution for each agent. The results of each step in capturing process will finally affect the AGR membership function to speed up the convergence of the competitive neural network. The experiments result show that this approach is more effective for the mobile agents to capture evaders.

show abstract

Multi-Agent Reinforcement Learning for Problems with Combined Individual and Team Reward

Sheikh

Bölöni

2020

2020 International Joint Conference on Neural Networks (IJCNN)

View full text Add to dashboard Cite

Learning Distributed Cooperative Policies for Security Games via Deep Reinforcement Learning

Cited by 4 publications

References 8 publications

Multi-Agent Reinforcement Learning for Problems with Combined Individual and Team Reward

Multi-Agent Reinforcement Learning for Problems with Combined Individual and Team Reward

A novel approach for multi-agent cooperative pursuit to capture grouped evaders

Multi-Agent Reinforcement Learning for Problems with Combined Individual and Team Reward

Contact Info

Product

Resources

About