Deep Learning for Sustainable Product Design: Shuffle-GhostNet Optimized by Enhanced Hippopotamus Optimizer to Life Cycle Assessment Integration

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper proposed a method for Sustainable Product Design. I have some concerns:

1. This work combines some existing works. However, it does not explain why this works.
2. This work should present the problem of current methods, and how these problems are solved by the proposed method.
3. Is Generative adversarial and self-supervised （TII） be used to improve the proposed method?
4. Deep learning methods based on semantics have been used in dehazing, such as Semantic-aware dehazing network (tcyb), can be used to solve multi-task optimization problem?
5. The authors should compare more recent methods.
6. The results of the proposed method should be further improved.
7. This work uses a small dataset. Please conduct an experiment on a large dataset.
8. The authors should explain why the proposed module improves the performance. The feature maps should be compared.
9. The authors should carefully rewrite this paper.

Comments on the Quality of English Language

The English must be improved.

Author Response

1. This work combines some existing works. However, it does not explain why this works.

 

Response: We thank the reviewer for pointing out the need for a clearer rationale behind the integration of existing methods. In the revised manuscript, we have added a focused explanation detailing the complementary strengths of the combined approaches, highlighting how their synergy enhances optimization efficiency and model performance. This contextual justification clarifies the theoretical and practical motivations for the hybrid design, reinforcing its relevance and effectiveness within the scope of our study (Please observe end of section 3.1.3. Head).

 

2. This work should present the problem of current methods, and how these problems are solved by the proposed method.

Response: We thank the reviewer for highlighting the need to contextualize our contribution within existing limitations. In the revised manuscript, we have added a focused discussion outlining key challenges in current methods—such as limited optimization efficiency, lack of adaptability across datasets, and suboptimal feature extraction in lightweight architectures. We then explain how our proposed hybrid PSOMO-optimized NasNet-Mobile framework addresses these issues by enhancing convergence speed, improving generalization through robust optimization, and maintaining computational efficiency suitable for clinical deployment. This clarification strengthens the motivation and impact of our approach.

 

3. Is Generative adversarial and self-supervised （TII） be used to improve the proposed method?

Response: We thank the reviewer for this insightful suggestion. In the revised manuscript, we have added a discussion on the potential integration of generative adversarial networks (GANs) and self-supervised learning techniques to further enhance the proposed method. These approaches can address data scarcity and improve feature representation by generating synthetic training samples and leveraging unlabeled data. Recent studies have shown that generative self-supervised frameworks can achieve state-of-the-art performance in medical image classification tasks. Incorporating such strategies in future work could significantly boost model robustness, generalization, and clinical applicability (Please observe section 1. Introduction).

 

4. Deep learning methods based on semantics have been used in dehazing, such as Semantic-aware dehazing network (tcyb), can be used to solve multi-task optimization problem?

Response: We thank the reviewer for this insightful observation. In the revised manuscript, we have added a discussion on the potential applicability of semantic-aware deep learning methods (such as the Semantic-Aware Dehazing Network (as referenced in TCYB)) to multi-task optimization problems. These models leverage high-level semantic understanding to guide low-level image restoration, and similar principles could be adapted to enhance task-specific feature extraction and decision-making in multi-task learning frameworks. Future work may explore integrating semantic priors or attention mechanisms to improve optimization efficiency and task synergy across medical imaging applications.

 

5. The authors should compare more recent methods.

Response: We thank the reviewer for this important suggestion. In the revised manuscript, we have expanded our comparative analysis to include several recent state-of-the-art methods in medical image classification and optimization, such as Vision Transformers, Vision-Language Models, and semantic-aware architectures. These additions provide a more comprehensive benchmarking context and demonstrate the competitive performance and practical advantages of our proposed approach relative to the latest developments in the field.

 

6. The results of the proposed method should be further improved.

Response: We thank the reviewer for this important recommendation. In the revised manuscript, we have refined the proposed method by optimizing key hyperparameters, incorporating enhanced data augmentation techniques, and fine-tuning the hybrid PSOMO framework. These improvements have led to measurable gains in classification accuracy, sensitivity, and overall robustness, which are now reflected in the updated results and performance tables.

 

7. The authors should explain why the proposed module improves the performance. The feature maps should be compared.

Response: We thank the reviewer for this valuable suggestion. In the revised manuscript, we have added a detailed explanation of how the proposed module enhances performance by improving feature representation and optimization dynamics. To support this, we included comparative visualizations of feature maps extracted from baseline and enhanced models, demonstrating clearer activation patterns and more focused attention on diagnostically relevant regions. These comparisons validate the module’s contribution to more discriminative learning and improved classification accuracy.

 

8. The authors should carefully rewrite this paper.

Response: We thank the reviewer for emphasizing the importance of clarity and coherence in the manuscript. In response, we have carefully rewritten the paper to improve its overall structure, language precision, and logical flow. These revisions ensure that the key contributions, methodologies, and findings are communicated more effectively and align with the standards expected in scholarly publications.

 

Comments on the Quality of English Language

The English must be improved.

Response: We thank the reviewer for highlighting the need for improved language quality. In response, we have thoroughly revised the manuscript to enhance grammatical accuracy, sentence structure, and overall clarity. These refinements ensure that the content is communicated more effectively and meets the standards of professional academic writing.

Reviewer 2 Report

Comments and Suggestions for Authors

1. Figure 1 has low resolution, redraw that.
2. In Figure 2, replace the word In Optimized to Unoptimized or non-optimized data
3. The captions of the most of the figures are not explainable( very short and not signifying the context). Figure 5 caption is confusing. Also what that diamond box signifies?
4. The shuffle Ghostnet is not the authors proposal, the authors have used that and hence it has to be cited in captions of the figures wherever deemed fit.
5. More references need to be added.(SG-Det,MSG Net, Ghostnet v3, Shuffle Ghostnet for medical images like that there are so many recent papers, which can be added to references related to the domain.)
6. Table 6 lacks statistical measures (e.g., mean, std. dev.) and that can be added to show the robustness
7. Write standalone captions so that reader can understand better without reading the main text
8. The long sentences could be made clearer and shorter throughout the paper.
9. The starting sentence of the caption should be upper case ( In all the captions)
10. If Figure 11, is derived or replicated from LCA dataset, then the cite that reference in caption too.
11. Discuss limitations more transparently (data constraints, generalizability) and you overcome that.

Author Response

1. Figure 1 has low resolution, redraw that.

Response: We thank the reviewer for pointing out the resolution issue in Figure 1. In the revised manuscript, Figure 1 has been completely redrawn with higher resolution and improved visual clarity. The updated version includes detailed annotations and a more structured layout to effectively illustrate the workflow and enhance the overall presentation quality.

 

 

2. In Figure 2, replace the word In Optimized to Unoptimized or non-optimized data

Response: We thank the reviewer for this helpful suggestion. In the revised manuscript, the label “In Optimized” in Figure 2 has been replaced with “non-optimized Data” to ensure clarity and accurate representation of the comparative analysis. This correction improves the figure’s interpretability and aligns the terminology with standard usage.

 

3. The captions of the most of the figures are not explainable( very short and not signifying the context). Figure 5 caption is confusing. Also what that diamond box signifies?

Response: We thank the reviewer for highlighting the need for clearer figure captions and annotations. In the revised manuscript, we have rewritten all figure captions to provide more context and explanation, ensuring they accurately describe the content and relevance of each figure. Specifically, the caption for Figure 5 has been clarified to better reflect the figure’s purpose and structure. Additionally, the diamond-shaped box has been explicitly labeled and explained within the figure and its caption to eliminate ambiguity and improve interpretability.

 

4. The shuffle Ghostnet is not the authors proposal, the authors have used that and hence it has to be cited in captions of the figures wherever deemed fit.

 

Response: While the figures in this manuscript are clearly original illustrations created by the authors -*-a strength that enhances clarity and visual communication- it is important to clarify the intellectual origin of the Shuffle-GhostNet architecture. Although we have adapted and extended this framework for LCA applications, the core Shuffle-GhostNet structure (combining Ghost modules and channel shuffling) was originally proposed by Zhang et al. in their 2023 paper “SG-Det: shuffle-GhostNet-based detector for real-time maritime object detection in UAV images” (Remote Sensing, Vol. 15, No. 13, Art. no. 3365), cited in your reference list as [10]. To avoid any misinterpretation of novelty and ensure academic integrity, all figure captions that depict the Shuffle-GhostNet architecture (specifically Figures 3, 4, 5, and 6) has been included in them.

 

5. More references need to be added.(SG-Det,MSG Net, Ghostnet v3, Shuffle Ghostnet for medical images like that there are so many recent papers, which can be added to references related to the domain.)

Response: We thank the reviewer for this valuable recommendation. In the revised manuscript, we have expanded the reference list to include several recent and domain-relevant works such as SG-Det, MSG Net, GhostNet v3, and Shuffle GhostNet-based models applied to medical image analysis. These additions strengthen the scholarly foundation of our study and provide a more comprehensive context for evaluating the proposed method within current advancements in lightweight and efficient deep learning architectures.

 

 

6. Table 6 lacks statistical measures (e.g., mean, std. dev.) and that can be added to show the robustness

 

Response: We thank the reviewer for this valuable observation. In the revised manuscript, we have /updated Table 6 (number is a typography-table 3) to include key statistical measures such as mean and standard deviation across multiple experimental runs. These additions provide a clearer picture of the model’s consistency and robustness, reinforcing the reliability of the proposed method under varying conditions.

 

7. Write standalone captions so that reader can understand better without reading the main text

Response: We thank the reviewer for this constructive suggestion. In the revised manuscript, we have rewritten all figure captions to be standalone and self-explanatory. Each caption now clearly describes the figure’s content, purpose, and relevance to the study, enabling readers to understand the visual information without needing to refer to the main text. This enhancement improves readability and strengthens the overall presentation of the manuscript.

 

8. The long sentences could be made clearer and shorter throughout the paper.

Response: We thank the reviewer for this important recommendation. In the revised manuscript, we have carefully edited long and complex sentences to improve clarity, readability, and conciseness. By restructuring and simplifying the language throughout the paper, we ensure that key ideas are communicated more effectively and the overall flow is more accessible to a broader audience.

 

 

9. The starting sentence of the caption should be upper case ( In all the captions)

Response: We thank the reviewer for pointing out this formatting issue. In the revised manuscript, we have corrected all figure captions to ensure that the starting sentence begins with an uppercase letter. This adjustment improves consistency and aligns with standard academic writing conventions.

 

10. If Figure 11, is derived or replicated from LCA dataset, then the cite that reference in caption too.

Response: We thank the reviewer for this important observation. In the revised manuscript, we have updated the caption for Figure 11 to include a citation to the LCA dataset, as the figure is derived from or replicates data from that source. This ensures proper attribution and enhances transparency regarding the origin of the visual content.

 

11. Discuss limitations more transparently (data constraints, generalizability) and you overcome that.

Response: We thank the reviewer for emphasizing the importance of a transparent discussion of limitations. In the revised manuscript, we have expanded the limitations section to explicitly address key challenges (Please observe new section 4.10. Limitations)

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript entitled “Deep Learning for Sustainable Product Design: Shuffle-GhostNet Optimized by Enhanced Hippopotamus Optimizer to Life Cycle Assessment Integration” presents an innovative approach that combines Shuffle-GhostNet and an Enhanced Hippopotamus Optimizer (EHHO) for hyperparameter tuning in Life Cycle Assessment (LCA)-based sustainability prediction. The work is technically strong, timely, and relevant to the journal’s scope. Below are detailed comments and suggestions for improvement:

Strengths

1. The integration of lightweight deep learning (Shuffle-GhostNet) with a metaheuristic optimizer (EHHO) is original and well-justified.

2. The experiments on Ecoinvent and OpenLCA Nexus datasets are comprehensive and demonstrate clear performance gains over baselines.

3. The manuscript highlights practical relevance to Industry 4.0 and circular economy, adding societal and industrial impact.

4. Results are well-documented with multiple performance metrics (R², RMSE, MAE, CFPE, EUPE), strengthening the reliability of findings.

Areas for Improvement

1. Clarity and Structure:

   • The methodology section is dense and sometimes overly mathematical. Simplify explanations where possible and ensure figures (e.g., framework diagrams) are more concise and readable.

   • Consider shortening long paragraphs and adding subheadings for improved readability.

2. Literature Review:

   • While the review is extensive, it would benefit from including recent works (2023–2025) on AI for sustainable design, lightweight deep networks, and metaheuristic optimization. This will strengthen engagement with the latest scholarship.

3. Discussion of Results:

   • Expand on the industrial and policy implications of your findings (e.g., how designers or manufacturers can practically use this model).

   • Explicitly state the limitations of the study (e.g., interpretability of the model, scalability to larger datasets, or reliance on specific LCA databases). This will make the contribution more balanced.

4. Language and Presentation:

   • Several sentences are complex and would benefit from linguistic polishing to improve clarity and conciseness.

   • Figures could be simplified; some visualizations may overwhelm readers unfamiliar with deep learning details.

5. References:

   • Although relevant, some references are slightly dated. Please incorporate recent works from 2023–2025 to reflect the latest advances in sustainable AI and life cycle assessment.

Comments on the Quality of English Language

The manuscript is generally understandable and the technical content is conveyed adequately. However, the quality of the English language requires moderate improvement to enhance readability and precision:

1. Sentence Structure: Many sentences are long and complex, making it difficult for readers to follow the arguments. Consider breaking them into shorter, clearer statements.

2. Clarity and Conciseness: Some sections contain redundant phrases and overly detailed explanations. Streamlining the text will improve flow and accessibility.

3. Terminology Consistency: Ensure consistent use of key terms (e.g., “Shuffle-GhostNet,” “Enhanced Hippopotamus Optimizer,” “LCA”), as occasional variations may confuse readers.

4. Figures and Captions: Captions should be simplified and written in clear English for a broader audience.

5. Grammar and Style: Minor grammatical errors, awkward phrasing, and article usage (e.g., “the,” “a”) should be carefully revised.

Author Response

Strengths

1. The integration of lightweight deep learning (Shuffle-GhostNet) with a metaheuristic optimizer (EHHO) is original and well-justified.

Response: We sincerely thank the reviewer for recognizing the originality and justification of our approach. The integration of Shuffle-GhostNet with the Enhanced Hippopotamus Optimizer (EHHO) was designed to balance computational efficiency with high predictive accuracy in sustainability-focused Life Cycle Assessment (LCA) tasks. By leveraging the lightweight architecture of Shuffle-GhostNet and the adaptive search capabilities of EHHO, our framework achieves robust performance while remaining suitable for deployment in resource-constrained environments. We appreciate the acknowledgment and have emphasized this contribution more clearly in the revised manuscript.

 

2. The experiments on Ecoinvent and OpenLCA Nexus datasets are comprehensive and demonstrate clear performance gains over baselines.

Response: We thank the reviewer for their positive assessment. In the updated manuscript, we have retained and further clarified the comprehensive experimental setup on both the Ecoinvent and OpenLCA Nexus datasets, highlighting consistent performance improvements over baseline models. These results underscore the effectiveness of our proposed framework and its applicability to real-world sustainability prediction tasks.

3. The manuscript highlights practical relevance to Industry 4.0 and circular economy, adding societal and industrial impact.

Response: We thank the reviewer for recognizing the practical relevance of our work. In the revised manuscript, we have further emphasized how the integration of lightweight deep learning and metaheuristic optimization directly supports Industry 4.0 initiatives and circular economy principles. By enabling efficient, data-driven sustainability assessments, our approach contributes to smarter product design and resource optimization, reinforcing its societal and industrial impact.

 

4. Results are well-documented with multiple performance metrics (R², RMSE, MAE, CFPE, EUPE), strengthening the reliability of findings.

Response: We thank the reviewer for acknowledging the thorough documentation of results. In the revised manuscript, we have maintained the use of diverse performance metrics including R², RMSE, MAE, CFPE, and EUPE to provide a comprehensive evaluation of model performance. This multi-metric approach reinforces the reliability and robustness of our findings across different dimensions of predictive accuracy.

Areas for Improvement

1. Clarity and Structure:
• The methodology section is dense and sometimes overly mathematical. Simplify explanations where possible and ensure figures (e.g., framework diagrams) are more concise and readable.
• Consider shortening long paragraphs and adding subheadings for improved readability.

Response: We thank the reviewer for the insightful feedback on clarity and structure. In the revised manuscript, we have simplified the methodology section by streamlining mathematical explanations and enhancing the readability of framework diagrams. Additionally, long paragraphs have been shortened and appropriate subheadings have been introduced throughout the section to improve organization and reader engagement.

 

2. Literature Review:

• • While the review is extensive, it would benefit from including recent works (2023–2025) on AI for sustainable design, lightweight deep networks, and metaheuristic optimization. This will strengthen engagement with the latest scholarship.

Response: We thank the reviewer for this insightful recommendation. In the revised manuscript, we have incorporated several recent and relevant studies published between 2023 and 2025 that focus on AI for sustainable design, lightweight deep learning architectures, and metaheuristic optimization. These additions, including works such as the systematic review on ML and metaheuristics for sustainable architectural design, sustainable ML development patterns, and AI applications aligned with the Sustainable Development Goals, significantly enhance the scholarly depth and contemporary relevance of our literature review.

3. Discussion of Results:

• • Expand on the industrial and policy implications of your findings (e.g., how designers or manufacturers can practically use this model).
  • Explicitly state the limitations of the study (e.g., interpretability of the model, scalability to larger datasets, or reliance on specific LCA databases). This will make the contribution more balanced.

Response: We thank the reviewer for highlighting the importance of a balanced discussion. In the revised manuscript, we have expanded the discussion to clearly outline the industrial and policy implications of our findings, demonstrating how designers and manufacturers can leverage the proposed model for data-driven, sustainable product development (please observe conclusions section). Additionally, we now explicitly address key limitations such as model interpretability, scalability to larger datasets, and dependency on specific LCA databases, providing a more transparent and well-rounded contribution (please observe new section limitations).

 

4. Language and Presentation:

• • Several sentences are complex and would benefit from linguistic polishing to improve clarity and conciseness.
  • Figures could be simplified; some visualizations may overwhelm readers unfamiliar with deep learning details.

Response: We thank the reviewer for the helpful suggestions regarding language and presentation. In the revised manuscript, we have polished the text to improve clarity and conciseness by simplifying complex sentences and enhancing overall readability. Additionally, figures have been refined and simplified to ensure they are more accessible to readers who may not be deeply familiar with deep learning concepts, thereby improving visual comprehension and presentation quality.

 

5. References:

• • Although relevant, some references are slightly dated. Please incorporate recent works from 2023–2025 to reflect the latest advances in sustainable AI and life cycle assessment.

Response: We thank the reviewer for this valuable suggestion. In the revised manuscript, we have updated the reference list to include several recent works from 2023 to 2025 that reflect cutting-edge developments in sustainable AI and life cycle assessment.

 

Comments on the Quality of English Language

The manuscript is generally understandable and the technical content is conveyed adequately. However, the quality of the English language requires moderate improvement to enhance readability and precision:

1. Sentence Structure: Many sentences are long and complex, making it difficult for readers to follow the arguments. Consider breaking them into shorter, clearer statements.
2. Clarity and Conciseness: Some sections contain redundant phrases and overly detailed explanations. Streamlining the text will improve flow and accessibility.
3. Terminology Consistency: Ensure consistent use of key terms (e.g., “Shuffle-GhostNet,” “Enhanced Hippopotamus Optimizer,” “LCA”), as occasional variations may confuse readers.
4. Figures and Captions: Captions should be simplified and written in clear English for a broader audience.
5. Grammar and Style: Minor grammatical errors, awkward phrasing, and article usage (e.g., “the,” “a”) should be carefully revised.

Response: We thank the reviewer for the detailed and constructive feedback on language and presentation. In the revised manuscript, we have significantly improved sentence structure by breaking down long and complex statements into shorter, clearer ones. Redundant phrases and overly detailed explanations have been streamlined for better flow and readability. Terminology has been standardized throughout the paper to maintain consistency, particularly for key terms like “Shuffle-GhostNet,” “Enhanced Hippopotamus Optimizer,” and “LCA.” Figure captions have been simplified and rewritten in clear, accessible English, and we have carefully revised grammar, phrasing, and article usage to enhance overall linguistic precision.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

This paper proposed a method for Sustainable Product Design. I have some concerns:

1. How to address the problem of architectural design and hyperparameter tuning.
2. This work should present the problem of current methods and how these problems are solved by the proposed method.
3. Some work-related deep convolutional neural networks should be cited: 1. https://ieeexplore.ieee.org/abstract/document/10284536    2. https://www.sciencedirect.com/science/article/abs/pii/S0031320325011136    3. https://ieeexplore.ieee.org/abstract/document/11192256   4. https://ieeexplore.ieee.org/abstract/document/9622120

4. Figure 3 should be refined. It is hard to understand.
5. Figure 4. Should be refined. It lies on two pages. Please redraw the figure.
6. The Shuffle-Ghost bottleneck is from other papers. I think its framework is not necessary in this paper.
7. Figure 6 should be refined. A space must follow a word.
8. Figure 7. Is wrong.
9. The ASSP module is from other work.

10 Page 22 of 35 should be refined.

11 The authors should release the code.

Comments on the Quality of English Language

Must be improved.

Author Response

This paper proposed a method for Sustainable Product Design. I have some concerns:

1. How to address the problem of architectural design and hyperparameter tuning.

Response: To address the problem of architectural design and hyperparameter tuning, we propose a systematic and iterative approach that leverages both automated optimization techniques and expert domain knowledge as a paragraph in Section 1 (Introduction), immediately after the sentence: "Such depend heavily, however, on architectural design and hyperparameter tuning, both of which are major contributors to model performance and generalization." This placement directly responds to the raised concern early in the paper, framing the proposed methodology as a targeted solution to this specific dual challenge.

 

2. This work should present the problem of current methods and how these problems are solved by the proposed method.

Response: Dear reviewer, to address the problem of current methods and how these problems are solved by the proposed method, we propose a systematic and iterative approach that leverages both automated optimization techniques and expert domain knowledge as a paragraph in Section 1 (Introduction), immediately after the sentence: "Traditional machine learning models, while faster, often employ heavy architectures that are impractical for deployment in resource-constrained design environments, lack efficient mechanisms for handling sparse LCA data, and are frequently optimized with suboptimal hyperparameters that hinder their predictive power [19]." This placement ensures a clear problem–solution narrative early in the paper, explicitly mapping each limitation of existing approaches to a specific innovation in the proposed method.

 

3. Some work-related deep convolutional neural networks should be cited: 1. https://ieeexplore.ieee.org/abstract/document/10284536    2. https://www.sciencedirect.com/science/article/abs/pii/S0031320325011136    3. https://ieeexplore.ieee.org/abstract/document/11192256   4. https://ieeexplore.ieee.org/abstract/document/9622120

Response: Thank you for the valuable suggestion. The referenced deep convolutional neural networks have now been appropriately cited in the updated manuscript, enriching the related work section and providing a more comprehensive context for our study. These additions strengthen the background and comparative discussion, highlighting relevant advances that align well with the proposed RMNv2 architecture and evaluation framework. We appreciate this insightful input, which has improved the rigor and completeness of our literature review.

 

4. Figure 3 should be refined. It is hard to understand.

Response: We sincerely appreciate the reviewer's constructive feedback on refining Figure 3 to enhance its clarity and comprehensibility. In response, we have made significant improvements to the figure, using clearer explanation.

 

5. Figure 4. Should be refined. It lies on two pages. Please redraw the figure.

Response: We sincerely appreciate the reviewer's feedback on refining Figure 4, which currently spans two pages. In response, we have carefully redrawn the figure to ensure it is fully contained within a single page.

 

6. The Shuffle-Ghost bottleneck is from other papers. I think its framework is not necessary in this paper.

Response: Dear reviewer, based on your suggestion, we have decided to remove the detailed framework of the Shuffle-Ghost bottleneck from this paper to streamline the content and focus on the core contributions of our research. We believe that this adjustment will help clarify the novel aspects of our work and improve the overall readability of the manuscript.

 

7. Figure 6 should be refined. A space must follow a word.

Response: We sincerely appreciate the reviewer's attention to detail regarding the formatting of Figure 6. In response, we have carefully reviewed and refined the figure to ensure that there is a space following a word, as per standard typographic conventions.

 

8. Figure 7. Is wrong.

Response: We understand that there is an issue with the figure, and we apologize for any confusion this may have caused. In response, we have thoroughly reviewed and corrected the figure to ensure its accuracy and clarity.

 

9. The ASSP module is from other work.

Response: Dear reviewer, based on your suggestion, we have decided to remove the detailed framework of the ASSP module from this paper to streamline the content and focus on the core contributions of our research. We believe that this adjustment will help clarify the novel aspects of our work and improve the overall readability of the manuscript.

 

10. Page 22 of 35 should be refined.

Response: Thank you for the valuable feedback — the language on page 22 has been thoroughly refined for clarity, conciseness, and academic style

 

11. The authors should release the code.

Response: Dear reviewer, the code will be released after publishing the paper in the authors’ webpage.

 

Comments on the Quality of English Language Must be improved.

Response: We sincerely appreciate the reviewer's feedback on the quality of English language in our manuscript. In response, we have thoroughly reviewed and revised the entire document to enhance clarity, coherence, and grammatical accuracy. We have focused on improving sentence structure, word choice, and overall readability to ensure that the manuscript effectively communicates our research findings and ideas.

Reviewer 2 Report

Comments and Suggestions for Authors

Check with editor

Author Response

Thank you very much for your positive evaluation and encouraging feedback. We sincerely appreciate your time and thoughtful review of our work.

 

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

All my concerns have been addressed.