Applications of Multi-Agent Deep Reinforcement Learning: Models and Algorithms

Abstract

Recent advancements in deep reinforcement learning (DRL) have led to its application in multi-agent scenarios to solve complex real-world problems, such as network resource allocation and sharing, network routing, and traffic signal controls. Multi-agent DRL (MADRL) enables multiple agents to interact with each other and with their operating environment, and learn without the need for external critics (or teachers), thereby solving complex problems. Significant performance enhancements brought about by the use of MADRL have been reported in multi-agent domains; for instance, it has been shown to provide higher quality of service (QoS) in network resource allocation and sharing. This paper presents a survey of MADRL models that have been proposed for various kinds of multi-agent domains, in a taxonomic approach that highlights various aspects of MADRL models and applications, including objectives, characteristics, challenges, applications, and performance measures. Furthermore, we present open issues and future directions of MADRL.

Federated Deep Reinforcement Learning for Prediction-Based Network Slice Mobility in 6 G Mobile Networks

“…9(c) and Fig. 9(d), when |N | ∈ [5,20], we can see that the Propose-MDP scheme has the highest average revenue and profit, followed by the proposed scheme with χ = 4, 3, 2, the NoPre-SA, NoPre-DRL, NoPre-Reset schemes, the proposed scheme with χ = 1, and finally the NoPre-Random scheme. Note that the performances of the Propose-MDP scheme and the NoPre-SA schemes significantly decrease with N , which demonstrates that the exhaustive search-based schemes are hard to deal with scenarios with large numbers of BSs.…”

“…For instance, the authors in [19] discussed the effect of noise on the observation of agents and tried to extract real and complete observations from the original ones. In [20] and [21], the authors studied the problem of unstable feedback of agents when dealing with highly dynamic environments. In addition, the combination of future information prediction and DRL has also been extensively studied [32].…”

Deep reinforcement learning classification of sparkling wines based on ICP-MS and DOSY NMR spectra

“…Deep learning using artificial neural networks (ANN) is increasingly used in the field of NMR with significant growth in the last few years ( Cobas, 2020 ). Different algorithms are applied for classification or regression problems ( Ibrahim et al, 2021 , Schartner et al, 2023 , Wang et al, 2020 ). The aim of this paper was to trace the geographical origin by using the deep learning classification of sparkling wines based on their ICP-MS and DOSY NMR spectra represented in the reduced space.…”

Multi-Agent Deep Reinforcement Learning Assisted Pre-connect Handover Management

“…MADRL extends the functions of DRL with MARL. MADRL enables multiple agents to interact with an environment to solve complex problems that the traditional DRL technique is not able to handle [10], particularly with distributed learning systems.…”

“…MARL is a group of agents that interact with the operating environment and interact with each other to achieve the goals [10]. MADRL extends the functions of RL and MARL with deep learning [10].…”