Particle Swarm-Based Federated Learning Approach for Early Detection of Forest Fires

Round 1

Reviewer 1 Report

Dear author(s),

Regarding the submission "Particle Swarm based Federated Learning Approach for Early Detection of Forest Fires", please find my comments below.

The article addresses a very important topic, proposing a model based on particle swarm and federated learning approach, able to detect forest fires as early as possible, but requires some revisions to make it suitable for publication. The low resolution and font (letter size and style) used in some figures is not readable (see Figure 4, for example).

The introduction presents a sufficient background, including related works relevant to the research. Moreover, the research design sounds appropriate, but the description of methods can be improved. The division between the training (1832 images) and test (68 images) sets is very unbalanced, 96.4% and 3.6% respectively. It is generally more common to use k-folds cross-validation. When hold-out validation is used, the common split is about 80% of the data for training and the remaining 20% of the data for testing. Therefore, this issue needs to be clarified by the authors.

Discussion of the results must be improved. The paper is very succinct, dedicating only a few brief paragraphs, but without details, and citing few or no references to discuss the results obtained. Considering that the journal aims to reach an international audience, and also as a way of valuing the analytical arrangement and the results, it is opportune to expand the discussion to include previous studies, in order to make it more evident how the findings of the present study are related to the results of previous studies.

From the foregoing, I recommend accepting the paper after review (corrections/clarifications) on these aspects.

Sincerely.

Author Response

Reviewer-1:

1.The article addresses a very important topic, proposing a model based on particle swarm and federated learning approach, able to detect forest fires as early as possible, but requires some revisions to make it suitable for publication. The low resolution and font (letter size and style) used in some figures is not readable (see Figure 4, for example).

Statement: As per the above mentioned query, changes to the font used in figures are changed. The resolution is also changed. Figure 4 is also changed.

2.The introduction presents a sufficient background, including related works relevant to the research. Moreover, the research design sounds appropriate, but the description of methods can be improved. The division between the training (1832 images) and test (68 images) sets is very unbalanced, 96.4% and 3.6% respectively. It is generally more common to use k-folds cross-validation. When hold-out validation is used, the common split is about 80% of the data for training and the remaining 20% of the data for testing. Therefore, this issue needs to be clarified by the authors.

Statement: As per the above mentioned query, the description is provided in 5.2.1

3.Discussion of the results must be improved. The paper is very succinct, dedicating only a few brief paragraphs, but without details, and citing few or no references to discuss the results obtained. Considering that the journal aims to reach an international audience, and also as a way of valuing the analytical arrangement and the results, it is opportune to expand the discussion to include previous studies, in order to make it more evident how the findings of the present study are related to the results of previous studies.

Statement: As per the above mentioned query, the changes are made and provided in section 5.

Reviewer 2 Report

This paper is of interest to use federated learning to estimate the bushfire. It is an important study to improve the resilience to the climate change and associated impacts, through early warning solution. Overall, the writing of this paper is good and the analysis is good. It has the potential to be accepted while some revisions are needed.

Abstract: It is too broad to present the key work and key findings of this paper. Please shorten the current information, and add more original findings.

Introduction:

Line 24-26, please add the time. You are not encouraged to present more about other challenges such  as Yellow fever. Just need to mention, the world is now challenged by many disasters, bushfire is one of them. Otherwise, it is not focus.

Iine 34-41, please add some information about the disaster management. Mitigation, adaptation and preparation, as well as the improvement is disaster management. Please refer: Beating the urban heat: Situation, background, impacts and the way forward in China. Renewable and Sustainable Energy Reviews, 161, 112350.

Line 42-82, it is not on the right way. Please do not introduce more about the FL. You are encouraged to describe how to improvement the management capacity, and what kinds of methods can be adopted. Then, FL is an emerging method to support this. The introduction of the FL can go to method section, for many algorithms have been developed. You need to select one algorithm based on sufficient justification.

Method section: It is not well written.

First, authors have presented the FL in section 2, but you need to present the ML methods, and make proper justifications. Please refer:

Mapping cation exchange capacity using a quasi-3d joint inversion of EM38 and EM31 data. Soil and Tillage Research, 200, 104618.

Clay content mapping and uncertainty estimation using weighted model averaging. Catena, 209, 105791.

Second, authors should present the case study and data collection, as well as the accuracy.

Third, authors have to present how to assess the accuracy. Please refer:

Mapping cation exchange capacity using a quasi-3d joint inversion of EM38 and EM31 data. Soil and Tillage Research, 200, 104618.

Clay content mapping and uncertainty estimation using weighted model averaging. Catena, 209, 105791.

Fourth, authors should present the training and testing samples.

Discussion:

First, authors have to discuss the method for bushfire detection, the strength and the shortcoming.

Second, authors should add a section of the uncertainty analysis.

Author Response

1.Abstract: It is too broad to present the key work and key findings of this paper. Please shorten the current information, and add more original findings.

statement: As per the above mentioned query, changed the Abstract with the current information.

2.Introduction:

Line 24-26, please add the time. You are not encouraged to present more about other challenges such  as Yellow fever. Just need to mention, the world is now challenged by many disasters, bushfire is one of them. Otherwise, it is not focus.

Statement: As per the above mentioned query, Added the time of Australian Bushfire. Removed the details about yellow fever. Mentioned about forest fires. Please refer to introduction section.

3.Iine 34-41, please add some information about the disaster management. Mitigation, adaptation and preparation, as well as the improvement is disaster management. Please refer: Beating the urban heat: Situation, background, impacts and the way forward in China. Renewable and Sustainable Energy Reviews, 161, 112350. Please refer to introduction section.

statement: As per the above mentioned query , added data related to disaster management,mitigation,adaption and preparation as well as is disaster management. Also referred to the paper you have mentioned. Please refer to section 1.1

4.Line 42-82, it is not on the right way. Please do not introduce more about the FL. You are encouraged to describe how to improvement the management capacity, and what kinds of methods can be adopted. Then, FL is an emerging method to support this. The introduction of the FL can go to method section, for many algorithms have been developed. You need to select one algorithm based on sufficient justification.---

statement: According to the above mentioned query ,Changed the complete Introduction part with all the specified requirements in section 1.

5.Method section: It is not well written.

First, authors have presented the FL in section 2, but you need to present the ML methods, and make proper justifications. Please refer:---Also referred the ML methods

Mapping cation exchange capacity using a quasi-3d joint inversion of EM38 and EM31 data. Soil and Tillage Research, 200, 104618.

Clay content mapping and uncertainty estimation using weighted model averaging. Catena, 209, 105791.

Statement: As per the above mentioned query, presented the ML method ad referred the papers. Please refer to section 2.1.

Second, authors should present the case study and data collection, as well as the accuracy.----

Statement: As per the above mentioned query, the details have been revised in the paper in the section 5.2

 Third, authors have to present how to assess the accuracy. Please refer:

Mapping cation exchange capacity using a quasi-3d joint inversion of EM38 and EM31 data. Soil and Tillage Research, 200, 104618.---

Clay content mapping and uncertainty estimation using weighted model averaging. Catena, 209, 105791.----

Statement:As per the above mentioned query, the accuracy is assessed in detail in section 5.4.

Fourth, authors should present the training and testing samples.----

Statement:As per the above mentioned query, provided the training and testing samples in section 5.2.

6.Discussion:

First, authors have to discuss the method for bushfire detection, the strength and the shortcoming.------

 Second, authors should add a section of the uncertainty analysis.

Statement:As per the above mentioned query, the mentioned details are available in section 5.4.

Reviewer 3 Report

1. Make your abstract specific to highlight key content areas, the purpose of your research, the relevance or significance of your work, and the preliminary results. It is a single, well-developed paragraph of approximately 250 words that is indented and single-spaced. The purpose of the abstract is to summarize all sections of the paper.

2. An acronym is a word or name formed from the initial components of a longer name or phrase. Acronyms are usually formed from the initial letters of words, as in FL, which has not been abbreviated in the paper except for the abstract part. 

3. For the introduction part 

The authors need to better explain the context of this research, including why the research problem is important. The introduction should clearly explain the key limitations of prior work that are relevant to this paper. Contributions should be highlighted more. It should be made clear what is novel and how it addresses the limitations of prior work. The authors should add an example to illustrate the problem definition. 

4. The reference format needs to be more consistent. Please check the format carefully and ensure it is consistent for all references. Both numbered [1] and APA style has been used (). 

5. The authors should explain the differences between the prior work and the solution presented in this paper.

6. The related work section needs to be longer. The authors should add a table that compares the key characteristics of prior work to highlight their differences and limitations. The authors may also consider adding a line in the table to describe the proposed solution.

7. Fig1,3,4 have yet to be cited in the text before the figures are used. Also, tables should be cited properly and explained in the text. 

8. Novelty is unclear.

Author Response

1. Make your abstract specific to highlight key content areas, the purpose of your research, the relevance or significance of your work, and the preliminary results. It is a single, well-developed paragraph of approximately 250 words that is indented and single-spaced. The purpose of the abstract is to summarize all sections of the paper.

Statement: As per the above mentioned query, the abstract has been changed properly.

2. An acronymis a word or name formed from the initial components of a longer name or phrase. Acronyms are usually formed from the initial letters of words, as in FL, which has not been abbreviated in the paper except for the abstract part. 

Statement: As per the above mentioned query, checked for the acronyms all over the paper.

3. For the introduction part 

The authors need to better explain the context of this research, including why the research problem is important. The introduction should clearly explain the key limitations of prior work that are relevant to this paper. Contributions should be highlighted more. It should be made clear what is novel and how it addresses the limitations of prior work. The authors should add an example to illustrate the problem definition. ---

Statement: As per the above mentioned query, made changes to the introduction part accordingly.

4. The reference format needs to be more consistent. Please check the format carefully and ensure it is consistent for all references. Both numbered [1] and APA style has been used (). 

Statement: As per the above mentioned query, checked the references and maintained consistency all over the paper.

5. The authors should explain the differences between the prior work and the solution presented in this paper

Statement: As per the above mentioned query, explained the differences in table 1.

6. The related work section needs to be longer. The authors should add a table that compares the key characteristics of prior work to highlight their differences and limitations. The authors may also consider adding a line in the table to describe the proposed solution

Statement: As per the above mentioned query, changed the related work section and added a table to represent the summary works.

7. Fig1,3,4 have yet to be cited in the text before the figures are used. Also, tables should be cited properly and explained in the text. 

Statement: As per the above mentioned query, the changes in the table citations has been made. 

8. Novelty is unclear

Statement: As per the above mentioned query, explained about the novelty in the  introduction part .

Reviewer 4 Report

n my opinion, the article can be published, but only after major changes. The article is interesting. Therefore, I propose:

improve the abstract,

Improve the introduction - conclude what is the purpose of the article, show the novelty, describe the structure.

No research methodology, no research questions can be found either.

I propose to describe the experiment in detail.

Now the reader has to guess at many things.

Conclusions need to be significantly expanded.

Author Response

1. In my opinion, the article can be published, but only after major changes. The article is interesting. Therefore, I propose:

Improve the abstract

Statement: As per the above mentioned query, the abstract has been improved accordingly.

2. Improve the introduction - conclude what is the purpose of the article, show the novelty, describe the structure.

Statement: As per the above mentioned query, made necessary changes to the introduction part.

3. No research methodology, no research questions can be found either.

Statement: As per the above mentioned query, mentioned the research methodology and research questions as a separate section 3.

4. I propose to describe the experiment in detail.

Statement: As per the above mentioned query, the experiment is explained in detail in section 5.

5. Now the reader has to guess at many things.

Conclusions need to be significantly expanded

Statement: As per the above mentioned query, conclusion subsection has been modified accordingly.

Reviewer 5 Report

This paper focuses on a particle swarm based federated learning approach for early detection of forest fires. However, some descriptions are not clear. Some revisions are necessary in the manuscript.

1. Please further explain what are the innovations of the PSO-enabled FL approach besides combining FL and PSO?

2. Please explain why should PSO and FL be combined? There are many similar optimization algorithms. Is it feasible to use other algorithms?

3. Figure 4 is part of Figure 2. Is it necessary to describe it again?

4. Please make sure all symbols are defined, including those in the algorithm table.

5. In the paper, authors have focused on proper planning and scheduling of relief operations. Different scheduling methods need to be compared to reflect the strengths of your work, which can refer to：

[a] IEEE Trans. Ind. Inf. vol. 18, no. 2, pp. 835-846. 2022

[b] Energy, vol. 133, pp. 380-387, 2017

[c] IEEE Trans. Power Systems, vol. 34, no. 1, pp. 152-162, 2019

Author Response

1. Please further explain what are the innovations of the PSO-enabled FL approach besides combining FL and PSO?

Statement: As per the above mentioned query, the necessary changes can be viewed in related work section, section 2.5.

2. Please explain why should PSO and FL be combined? There are many similar optimization algorithms. Is it feasible to use other algorithms.

Statement: As per the above mentioned query, the necessary changes can be viewed in related work section, section 2.5.

3. Figure 4 is part of Figure 2. Is it necessary to describe it again?

Statement: As per the above mentioned query, we modified the figure 2.

4. Please make sure all symbols are defined, including those in the algorithm table.

Statement: As per the above mentioned query, defined all the symbols in table 2.

5. In the paper, authors have focused on proper planning and scheduling of relief operations. Different scheduling methods need to be compared to reflect the strengths of your work, which can refer to：

 [a] IEEE Trans. Ind. Inf. vol. 18, no. 2, pp. 835-846. 2022

[b] Energy, vol. 133, pp. 380-387, 2017

[c] IEEE Trans. Power Systems, vol. 34, no. 1, pp. 152-162, 2019

Statement: As per the above mentioned query, referred to the papers.

Round 2

Reviewer 1 Report

Dear authors,

The amendments made here have made the manuscript much better in many areas. However, the work remains very succinct in the discussion section (5.5), dedicating only a few brief paragraphs, but without details, and citing few or no references to discuss the results obtained. It is necessary to broaden the discussion to include previous studies in order to make more evident how the findings of the present study relate to the results of previous studies. It is worth mentioning that this problem remains from the first revision.

Author Response

he amendments made here have made the manuscript much better in many areas. However, the work remains very succinct in the discussion section (5.5), dedicating only a few brief paragraphs, but without details, and citing few or no references to discuss the results obtained. It is necessary to broaden the discussion to include previous studies in order to make more evident how the findings of the present study relate to the results of previous studies. It is worth mentioning that this problem remains from the first revision.

Statement: According to the above mentioned query, included the previous studies in table 5.Also, included the references of the previous studies also.

Reviewer 2 Report

Authors have well addressed all my concerns.

Author Response

We thank the reviewer for the positive comments.

Reviewer 3 Report

The authors have answered my concerns.

Author Response

We thank the reviewer for the positive comments.

Round 3

Reviewer 1 Report

I'm sorry, but the article still remains very succinct in the discussion section (5.5), citing only a single reference [60] to discuss the results obtained. I reinforce that it is necessary to expand this section (5.5) to include previous studies in the discussion of the article, in order to make it more evident how the findings of the present study are related to the results of previous studies. It is worth mentioning that this problem persists since the first revision.

Regards

Author Response

I'm sorry, but the article still remains very succinct in the discussion section (5.5), citing only a single reference [60] to discuss the results obtained. I reinforce that it is necessary to expand this section (5.5) to include previous studies in the discussion of the article, in order to make it more evident how the findings of the present study are related to the results of previous studies. It is worth mentioning that this problem persists since the first revision.

Statement: According to the above mentioned query, we elaborated  section 5.5 and included the previous studies to make it evident, how the findings of the present study are related to the results of previous studies..