All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
Authors have addressed all the comments from the reviewers, Hence, it is recommended to accept this paper in its current form.
[# PeerJ Staff Note - this decision was reviewed and approved by Yilun Shang, a PeerJ Section Editor covering this Section #]
The new version addressed most of my concerns.
The new version addressed most of my concerns.
The new version addressed most of my concerns.
no comment
The authors have addressed my comments and i suggest accepting the paper
no comment
The authors have addressed my comments and i suggest accepting the paper
no comment
The authors have addressed my comments and i suggest accepting the paper
no comment
The authors have addressed my comments and i suggest accepting the paper
Major revision is required for this paper based on lacking of some technical details in the presented work. Authors should also consider the following in the revised manuscript.
1. In "Abstract", authors should provide the quantitative performance achievement of the proposed method.
2. Carefully review the literature to cover the state-of-the-art works.
3. The details of why certain settings were selected should be clearly explained in their experiments.
[# PeerJ Staff Note: It is PeerJ policy that additional references suggested during the peer-review process should *only* be included if the authors are in agreement that they are relevant and useful #]
This paper proposed a Transformer-based Multi-Agent Communication (TMAC) method for partially observable multiagent cooperation. The writing is good, and the figures and tables are clear. Moreover, the reviewer found that the self-message fusion algorithm is novel.
Points to improve:
1. some claims are not suported by the experiments, for example, the abstract writes that "TMAC improves agents extracting features and generating output messages", but experiments do not contains any extracting features analysis.
2. many claims are not properly citated with previous works. For example, the authors say that "Full communication may bring a large amount of redundant information", in fact, previous works have verified this. It is better to cite previous works.
3. Literature review is not sufficient. For example, early-era deep MA communication methods (e.g., ACCNet, attention-based MA communication methods, and MA communication with limited-bandwidth are not mentioned.
4. Experiment setting can be improved. For example, including Transformer-based Multi-Agent cooperation methods and testing on more environments
Overall, the experiments design are good, including main results and ablation study.
Points to improve:
1. including Transformer-based Multi-Agent cooperation methods since your method is a Transformer-based one.
2. testing on more environments (e.g., SMAC). Currently, only testing on a grid environment.
3. some key claims can be suported by the experiments, for example, the abstract writes that "TMAC improves agents extracting features and generating output messages", but experiments do not contains any extracting features analysis.
no comment
The paper introduces TMAC, a Transformer-based communication algorithm for multi-agent systems under partial observability. It aims to enhance feature extraction and message generation in communication-restricted environments. It proposes a self-message fusion module to integrate information from various sources, thereby facilitating more effective coordination among agents. Please see below my comments:
5. Authors doesn’t explain why certain settings, like the number of Transformer layers, were chosen.
6. More thought on how TMAC manages message passing would be helpful, especially if many agents are involved at once, as indicated by my previous comment.
3. Transformers are known to use a lot of computing power, especially with many agents involved. It would be better if the paper indicates whether TMAC’s approach can work in settings that don’t have as many resources or need to run on simpler devices.
4. Since the paper only uses a single grid simulation for testing, it’s hard to tell if TMAC will work in other setups. Please clarify.
1. Since the paper addresses an important topic related to effective communication and sensing, the paper should be related to upcoming 6G generation of wireless. In particular, in multi-agent systems where agents need to sense their surroundings and communicate with each other simultaneously, 6G's ISAC [REF01] technology allows for seamless integration of both functions over shared infrastructure. This is especially relevant for partially observable environments in MARL, where sensing is as crucial as communication for agents to build an understanding of the environment, share situational awareness, and coordinate actions.
2. The paper can discuss more about the practical difficulties of using TMAC with real-world systems.
7. The conclusion may be enhanced.
References
[REF01] A. Bazzi and M. Chafii, “On Outage-Based Beamforming Design for Dual-Functional Radar-Communication 6G Systems,” in IEEE Transactions on Wireless Communications, vol. 22, no. 8, pp. 5598-5612, Aug. 2023, doi: 10.1109/TWC.2023.3235617
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.