Assessing the Applicability of Machine Learning Models for Robotic Emotion Monitoring: A Survey
, Marat Rostov 2, Jessica Sharmin Rahman 2,3, Khandaker Asif Ahmed 3,* and Md Zakir Hossain 2,3,*Round 1
Reviewer 1 Report
This paper summarizes the current work related to emotion recognition in robotics, which is very meaningful. There are the following comments hopefully to be adopted. (1) Section background summarizes the related works on sensors that monitor physiological signals of the body and does not summarize the work on emotion recognition with respect to facial expression recognition, gait, body movement, and speech. (2) This manuscript presents discrete and valence-arousal models for emotion. As far as we know there is another hierarchical model. We suggest that the authors introduce it. (3) In Section 3, a detailed description of the research approach is given, but the reader would prefer to get from this section a detailed summary about the research paper. (4) For Figures 5 and 6, we suggest that replacing with a table would make it clearer to the reader. (5) The authors should check the formatting of “arousal-valence” or “valence-arousal” throughout the text, for example line 55, line192. (6) The author should standardize the citation of the figures in the manuscript, for example line 297, line 313. (7) The author should also check some writing errors, for example, what is the meaning of 2018 in line 177?
Author Response
Please see the attachment
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper considers analyzing and determining which machine learning methods and signal sources are the most appropriate for emotion monitoring through robots. It surveyed existing works till August 2022 and provided a comparative analysis for discrete and arousal-valance emotions. The topic is interesting, but I have the following comments:
1. The Abstract should briefly state the general aspects of the review subject and main conclusions, including the principal objectives and scope of the review. I think the current Abstract cannot act as a summary of the information in this review paper.
2. Generally, the fundamental rationale of writing a review article is to make a readable synthesis of the best literature sources on a crucial research inquiry or a topic. Therefore, you should state the question to be dealt with and the motivation behind writing this review article. In lines 42-45, you claimed, “Currently, numerous ML methods and literature reviews showed the potentiality of different sensors to monitor human emotions, but there is still a lack of studies to identify suitable sources and ML models for robotic applications.” I think you should describe the central question more reasonably.
3. In lines 58-59, you state, “This paper aims to analyze and determine which machine learning methods and signal sources are the most appropriate for emotion monitoring through robots.” However, what is the result? Perhaps you should answer this question in the Conclusion section.
4. Also, please set boundaries for your work and explain why. I think the boundaries of topics are not clearly described in the current version.
5. The organization needs to improve. For example, in the Background section, you should explain why the sub-sections include Human Signal Sources, ML Models, etc.
6. In discussing the data, for example, in figure 5, you should List and define all variables for which data were sought and any assumptions made.
7. In the Conclusion, you should summarize the main findings, including the strength of evidence for each primary outcome.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
The authors reviewed emotion monitoring using machine learning models. The presentation of the article is well-organized. However, there are a few concerns in the article that need to be addressed for quality improvement.
Comments
1) In the introduction section, the authors have presented the previous studies. However, the authors have not addressed the research and limitations in the previous studies. In addition to this, the contribution and organization of the study need to be discussed.
2) The tabular form must be included in section 4 for discussing the accuracy results of different machine learning models. The same thing needs to be included in section 2
3) A new section needs to be created before the conclusion section to discuss the overall analysis of the review. In addition to this, future research directions and limitations of the current study need to be presented.
4) The type of the study is written as “Article”, as the content in the article is addressed as a review
Author Response
Please see the attachment
Author Response File: Author Response.pdf
