Multi-Objective Combinatorial Optimization Algorithm Based on Asynchronous Advantage Actor–Critic and Graph Transformer Networks
, Ming Cao 1, Wenbin Hu 1,*, Jing Sun 2, Hui Li 1, Yichen Wang 1, Weijie Zhou 1, Tiancheng Yin 1 and Ran Qian 1
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
Comments for author File: 
Comments.pdf
Author Response
Thank you sincerely for taking the time to read the manuscript.
Reviewer 2 Report
Comments and Suggestions for AuthorsThis paper proposes a novel multi-objective combinatorial optimization algorithm based on DRL. The algorithm has a good performance in terms of training efficiency and other aspects. Here are my suggestions：
1.Does the use of the A3C network actually reduce the variance and avoid locally optimal solutions?
2.Are the considerations in the simulation results comprehensive? Are there other performance metrics not considered?
3.Is the method applicable to complex multi-objective situations?
4.What is the core innovation and core work of this paper?
5.In the paper, the author focuses on using artificial intelligence methods in multi-objective optimization., and different artificial intelligence algorithms can be compared and analyzed and to show your advantages, which can refer to:
[a] IEEE Transactions on Industrial Informatics, DOI: 10.1109/TII.2024.3390595
[b] IEEE Transactions on Industrial Informatics, vol. 19, no. 11, pp. 10751-10762, 2023
[c] IEEE Transactions on Smart Grid, vol. 12, no. 6, pp. 4834-4842, Nov. 2021
[d] IEEE Transactions on Smart Grid, vol. 10, no. 2, pp. 1889-1897, March 2019
Comments on the Quality of English LanguageA proofreading is needed.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe paper presents a current and interesting topic related to multi-objective combinatorial optimization problems. Despite the fact that the introduced algoritm is adequately decribed and compared with other multi-objective optimization algoritms, a few observations have been made below that could help improve the quality of the paper.
1) The introduction section is not sufficient. The literature review is done quite vaguely. Clubbing of more than one reference article by one single statement for citation should be avoided. The authors neither discuss nor compare the issues presented in the references - lines 35, 40, 53, 56 and 111.
2) Figure 8, Figure 9, Figure 10 - Axes value labels are unreadable. Please increase the font size.
3) Line 2 - There is a word "Title" in the manuscript title.
4) Line 76 - The introduction chapter title is repeated in the text.
5) Line 150 - The Figure 1 title is not sufficient.
6) Lines 325-341 - A fair fragment of the template was left in the text.
7) Line 466 - Patents chapter is empty.
8) The Authors Contributions part is missing.
9) The list of abbreviations would be desirable.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 4 Report
Comments and Suggestions for AuthorsThis study introduces a novel multi-objective combinatorial optimization algorithm based on DRL. The topic is interesting. However, followings are the comments from the reviewer.
1). Some abbreviations have not been defined in the main text (such as MOCOP, DRL, GTN, etc. ).
2). The axis labels on results figures are too small. Try to use consistent font sizes in all figures.
3). Improve the paragraph breaking through the paper.
4). Some insights on the complexity of the algorithms should be provided.
Comments on the Quality of English LanguageIn general, the paper is well written. However, the use of English could be further improved.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Comments and Suggestions for AuthorsThe Authors have adressed all the suggestions and comments in the revised version.
There remains only a minor editing issue. According to a template, "Authors contributions" part should be listed using Authors' initials.
