Deep Learning and Machine Learning, Better Together Than Apart: A Review on Biometrics Mobile Authentication
Round 1
Reviewer 1 Report
(1) This paper is well written and provides a good review of previous studies. However, it is not clear what research questions you are trying to achieve through this literature review and what new knowledge you are trying to provide to the literature. Additionally, please comment further on the excellence of this paper.
(2) Please introduce in more detail the criteria for finding and classifying prior studies. Also, please provide a concise summary of the preceding studies mentioned in this study.
(3) A more visible and logical explanation of Table 6 is needed. Also, please supplement the interpretation and discussion of the review results. Also, please supplement your conclusions with implications and insights.
Author Response
Comment 1: This paper is well written and provides a good review of previous studies. However, it is not clear what research questions you are trying to achieve through this literature review and what new knowledge you are trying to provide to the literature. Additionally, please comment further on the excellence of this paper.
Response: We have updated the final section of our introduction to address what contributions we hope to achieve with our manuscript. We have updated a section to include a description of research questions and how we have achieved them through our review.
Comment 2: Please introduce in more detail the criteria for finding and classifying prior studies. Also, please provide a concise summary of the preceding studies mentioned in this study.
Response: We have made changes and have referenced studies to classify it with different biometrics authentication categories.
Comment 3: A more visible and logical explanation of Table 6 is needed. Also, please supplement the interpretation and discussion of the review results. Also, please supplement your conclusions with implications and insights.
Response: Our table 6 has been updated with a subjects section and elaborated upon in a paragraph explaining the main takeaways. We have elaborated on the implications and insights of our work in the conclusion and future work sections.
Reviewer 2 Report
In this paper, authors present a comprehensive review of recent findings related to the most popular biometric methodologies with ML and DL algorithms with the aim of guiding and informing researchers in their experiments related to biometric-based mobile security. I think this review on biometric authentication using machine learning and deep learning techniques is interesting, well written, organized, and clear, but it does not present enough novelty. In addition, the manuscript has some drawbacks that need to be addressed before a possible publication:
1.- Lines 341 and 391 show low quality images, I suggest the authors attend this for a better presentation.
2.- Regarding acronyms, I consider that authors should include the description or meaning at least the first time they are introduced in a text to give the reader a better understanding, for example, the acronym EER.
Minor editing of English language is required
Author Response
Comment 1: In this paper, authors present a comprehensive review of recent findings related to the most popular biometric methodologies with ML and DL algorithms with the aim of guiding and informing researchers in their experiments related to biometric-based mobile security. I think this review on biometric authentication using machine learning and deep learning techniques is interesting, well written, organized, and clear, but it does not present enough novelty. In addition, the manuscript has some drawbacks that need to be addressed before possible publication.
Response: Our manuscript presents novelty in the focused realm of study by showcasing many different dynamics across biometric mobile authentication within current and recent research findings in a comprehensive manner, allowing for a significant collection of information and advised instruction to be accessible in one place. We have updated the final section of our introduction to address what novelty our manuscript provides.
Comment 2: Lines 341 and 391 show low quality images, I suggest the authors attend this for a better presentation.
Response: We have replaced the images mentioned with higher quality, clearer pictures.
Comment 3: Regarding acronyms, I consider that authors should include the description or meaning at least the first time they are introduced in a text to give the reader a better understanding, for example, the acronym EER.
Response: Our manuscript has been reviewed an updated to include proper handling of acronyms.
Reviewer 3 Report
1- Figures should be made based on the authors' forecast horizon.
2- The discussion section in the paper should be revised and rewritten more strongly.
3- Table 6 should be more complete and the relevant database and the number of subjects should be mentioned.
4- In the discussion section, how is the future of such an approach predicted in the world?
Minor editing of English language required
especially : 1- Introduction and 2- Discussion
Author Response
Comment 1: Figures should be made based on the authors' forecast horizon.
Response: The figures that have been cited in this paper are important to give depth in knowledge of the categories discussed. They are helpful for readers to understand the context and the proof of concept for the given categories, touch, motion, keystroke, gait, facial, and ocular dynamics for biometric authentication. Other figures included also give an in depth look at the structure of algorithms used in specific studies that could not be efficiently described. For example, figure 5 presents a visualization of the keystroke data provided by different individuals in study 62. The figure visually explains how the use of advanced algorithms such as Principle Component Analysis (PCA) with biometric data can represent an individual in a unique manner, allowing for authentication. This figure provides a proof of concept for the validity of biometric authentication and gives context for how biometric data can work with algorithms to represent individuals. In figure 7, a human gait sequence is visualized from study 70. This figure presents how researchers may observe patters in a human gait cycle to properly differentiate and quantify segments of gait data. This figure provides context on how gait data can be collected and extracted to create a model based on user walking patterns. Figure 8 provides a pie graph depicting the proportion of architectures used in the facial dynamic authentication studies reviewed. This figure allows for a visual representation of a significant pattern observed in the types of algorithms used within facial dynamics, which is signified and elaborated on in the discussion and conclusion sections of the paper.
Comment 2: The discussion section in the paper should be revised and rewritten more strongly.
Response: We have made the necessary changes.
Comment 3: Table 6 should be more complete and the relevant database and the number of subjects should be mentioned.
Response: We have updated table 6 with a section depicting the databases used in each study with the number of subjects involved.
Comment 4: In the discussion section, how is the future of such an approach predicted in the world?
Response: We have elaborated on the future of this work in both the conclusion and future work sections.
Round 2
Reviewer 3 Report
Figures should be made based on the authors' forecast horizon.
Moderate editing of English language required
Author Response
Thank you very much for your valuable time for reviewing our manuscript.
1] Figures should be made based on the authors' forecast horizon.
We have included three new forecasting figures developed by authors.
Figure 10. Depiction of a hybrid architecture.
Figure 11. Depiction of the ideal behavioral biometric model.
Figure 12. Depiction of the ideal physiological biometric model.
Author Response File: 
Author Response.pdf
