A Survey System for Artificial Intelligence-Based Painting Using Generative Adversarial Network Techniques
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThank you for your work. Please find suggestions for areas of improvement below:
A clear description of what is meant by the term software may first need to be explained and the ways in which AI is utilized should be specified. The same holds for the characterization of the evaluation system.
It is not clear in the abstract what is being optimized and the rationale and theory behind it. More importantly the gap and need that has led to this research is not justified.
It is known that a generative adversarial network (GAN) is a deep learning architecture. It trains two neural networks to compete against each other to generate more authentic new data from a given training dataset. More explanation of what these two neural networks and why and how they are chosen are needed.
This research is intended to address three broad goals which are somewhat linked and cofounded with one another, namely:
“（1）To optimise the AI painting software based on generative adversarial network
technology in terms of functionality, ease of use, system performance and security, so as to make it more rich in functionality, more convenient in operation, more stable in performance and more reliable in security.
（2）To establish an evaluation system for AI painting software based on generative adversarial network technology to ensure that the software can fully integrate the four requirements of functionality, ease of use, system performance and security.
（3）To rank the functionality, ease of use, system performance and security of the software in order of importance through further research, which will help companies to invest more R&D time in the more important aspects.”
The objectives and outcomes may hence be subject to the unfolding of the above questions that seem to be missing a recognition and evaluation of alternatives.
Author Response
A clear description of what is meant by the term software may first need to be explained and the ways in which AI is utilized should be specified. The same holds for the characterization of the evaluation system.
Response:
We apply PyTorch software, which is a deep learning framework for neural networks developed by Facebook and widely used in both academia and industry. The GAN of the neural network is applied to this paper for AI drawing.
It is not clear in the abstract what is being optimized and the rationale and theory behind it. More importantly the gap and need that has led to this research is not justified.
Response:
In this paper, 123 indicators were initially established, and 47 indicators were optimised through exploratory factor analysis, and the importance of the indicators was ranked through fuzzy hierarchical analysis. Due to the limited length of the abstract, the authors can only briefly present the results of this paper .
It is known that a generative adversarial network (GAN) is a deep learning architecture. It trains two neural networks to compete against each other to generate more authentic new data from a given training dataset. More explanation of what these two neural networks and why and how they are chosen are needed.
Response:
A Generative Adversarial Network (GAN) consists of a generative model G and a discriminative model D. The purpose of a GAN is to learn the distribution of the training data, and in order to learn the distribution, first an input noise variable is defined, and next it is mapped to the data space, where G is a generative model consisting of a multilayer perceptual network with as parameters. In addition, a discriminative model is defined which is used to determine whether the input data is from the generative model or the training data, and the output of D is the probability that x is the training data. Finally, D is trained to judge the source of the data as accurately as possible, and G is trained to generate data that matches the distribution of the training data as closely as possible. The description of the two neural networks can be found in Refs. 21and 32.
This research is intended to address three broad goals which are somewhat linked and cofounded with one another, namely:
“（1）To optimise the AI painting software based on generative adversarial network
technology in terms of functionality, ease of use, system performance and security, so as to make it more rich in functionality, more convenient in operation, more stable in performance and more reliable in security.
（2）To establish an evaluation system for AI painting software based on generative adversarial network technology to ensure that the software can fully integrate the four requirements of functionality, ease of use, system performance and security.
（3）To rank the functionality, ease of use, system performance and security of the software in order of importance through further research, which will help companies to invest more R&D time in the more important aspects.”
The objectives and outcomes may hence be subject to the unfolding of the above questions that seem to be missing a recognition and evaluation of alternatives.
Response:
Based on the above objectives, this paper provides the following solutions as follows: the first objective is to construct 123 indicators based on the four aspects of functionality, ease of use, system performance, and security through Delphi, which is optimised by exploratory factor analysis; the second objective is to screen out the most important 47 indicators in order to construct the evaluation system of this paper; the third objective is to construct the evaluation system of this paper through fuzzy hierarchical analysis and fuzzy comprehensive evaluation, the four first-level indicators of functionality, ease of use, system performance, and security are ranked in order of importance, and the conclusion that functionality is the most important is drawn.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsComments and Suggestions for Authors
The aim of this paper is to construct an evaluation system for AI painting software based on generative adversarial network (GAN) technology, which optimizes the performance of the related software in terms of functionality, ease of use, system performance, and safety. The construction of the evaluation system involves synthesizing the results of processed data, which were obtained based on consultations with experts. The discussed topic is of great importance in the field of AI, as it may impact the development of creative technologies.
The article presents an interesting approach to analyzing the functionality and performance of AI software. A strong aspect of the paper is the use of advanced analytical methods (exploratory factor analysis and hierarchical analysis). The topic holds significant potential and is well justified from a literature perspective.
Abstract: This section presents a general outline of the research; however, it lacks focus on the proposed assumptions in lines 61-70. It is also recommended to include specific conclusions regarding the impact on software development and to describe how the proposed system could contribute to improving the software's functionality.
Introduction: This section provides a good introduction to the topic of AI and GAN. However, I recommend a more detailed discussion of existing technologies and the challenges related to their functionality. An expansion of this topic could be included in a new paragraph after line 51.
Research Objectives: The objectives are clearly formulated but should be presented in a more systematic manner. Research objectives are often supplemented by research hypotheses, which are assumptions that the authors aim to verify. The article lacks clearly defined hypotheses or research questions that could focus the study on more specific issues.
Methodology: The description of the methodology is detailed and well-organized. I particularly appreciate the use of advanced statistical methods, such as hierarchical methods and factor analysis.
Results and Discussion: The results of the study are presented clearly, but there is a lack of a more detailed discussion of their practical implications. The authors suggest that their evaluation system will help technology companies optimize software, but they do not provide specific examples of how organizations can use this in practice.
The article does not offer a deeper comparison of the results with other studies on AI technologies, including GAN-based systems. The discussion of the results is mainly limited to their presentation, without addressing how these findings relate to previous studies or how they may influence future work in this field.
Conclusions: The final conclusions are formulated concisely, but there is a lack of discussion on future research directions and potential improvements in the proposed system.
The article is valuable and contains a solid theoretical foundation and a clear methodology. However, several improvements are recommended to strengthen the clarity of the message and provide a fuller picture of the applied techniques and research outcomes. It is especially important to expand the discussion of the results and to add a broader comparative perspective.
Technical remarks:
Table 2. The caption is incomplete, which may make it difficult for readers to understand the content of the table. It is recommended to complete the description to clearly explain what the table contains and what the data represent.
It is worth ensuring that key terms, such as ‘Generative Adversarial Networks (GAN)’ and ‘AI painting software,’ are consistently used throughout the paper. In several places, these terms may be used somewhat variably, which could confuse readers.
Comments on the Quality of English Language
no comments
Author Response
Comments and Suggestions for Authors
The aim of this paper is to construct an evaluation system for AI painting software based on generative adversarial network (GAN) technology, which optimizes the performance of the related software in terms of functionality, ease of use, system performance, and safety. The construction of the evaluation system involves synthesizing the results of processed data, which were obtained based on consultations with experts. The discussed topic is of great importance in the field of AI, as it may impact the development of creative technologies.
The article presents an interesting approach to analyzing the functionality and performance of AI software. A strong aspect of the paper is the use of advanced analytical methods (exploratory factor analysis and hierarchical analysis). The topic holds significant potential and is well justified from a literature perspective.
Abstract: This section presents a general outline of the research; however, it lacks focus on the proposed assumptions in lines 61-70. It is also recommended to include specific conclusions regarding the impact on software development and to describe how the proposed system could contribute to improving the software's functionality.
Response:
This section presents the findings of the study, but also provides an explanation of the concerns with the hypotheses presented in lines 61-70. The first objective is to construct 123 indicators based on the four aspects of functionality, ease of use, system performance, and security through Delphi, which is optimised by exploratory factor analysis; the second objective is to screen out the most important 47 indicators in order to construct the evaluation system of this paper; the third objective is to construct the evaluation system of this paper through fuzzy hierarchical analysis and fuzzy comprehensive evaluation, the four first-level indicators of functionality, ease of use, system performance, and security are ranked in order of importance, and the conclusion that functionality is the most important is drawn.In addition, we include specific findings about the impact on the four aspects of software development of functionality, ease of use, system performance, and security.
Introduction: This section provides a good introduction to the topic of AI and GAN. However, I recommend a more detailed discussion of existing technologies and the challenges related to their functionality. An expansion of this topic could be included in a new paragraph after line 51.
Response:
This evaluation system is based on the latest advances in GAN. For example, if a company learns that the latest GAN technology can achieve automatic image enhancement through this evaluation system, it can study BIGGAN, an image generation model with excellent performance in image enhancement, and apply it to its newly developed software. After that, the company can also compare the images generated by the new software with the old software by using image evaluation methods such as LPIPS and FID. Some of the applications we have expanded in the new paragraph after lines 51-61.
Research Objectives: The objectives are clearly formulated but should be presented in a more systematic manner. Research objectives are often supplemented by research hypotheses, which are assumptions that the authors aim to verify. The article lacks clearly defined hypotheses or research questions that could focus the study on more specific issues.
Response:
The hypotheses of the three objectives have been verified by the authors. The first objective is to construct 123 indicators based on the four aspects of functionality, ease of use, system performance, and security through Delphi, which is optimised by exploratory factor analysis; the second objective is to screen out the most important 47 indicators in order to construct the evaluation system of this paper; the third objective is to construct the evaluation system of this paper through fuzzy hierarchical analysis and fuzzy comprehensive evaluation, the four first-level indicators of functionality, ease of use, system performance, and security are ranked in order of importance, and the conclusion that functionality is the most important is drawn. In addition, we include specific findings about the impact on the four aspects of software development of functionality, ease of use, system performance, and security.
Methodology: The description of the methodology is detailed and well-organized. I particularly appreciate the use of advanced statistical methods, such as hierarchical methods and factor analysis.
Response:
Thank you to the reviewers for their attention.
The statistical methods used in this paper, such as hierarchical methods and factor analyses, are designed to reduce the number of indicators as well as to rank the importance of the indicators.
Results and Discussion: The results of the study are presented clearly, but there is a lack of a more detailed discussion of their practical implications. The authors suggest that their evaluation system will help technology companies optimize software, but they do not provide specific examples of how organizations can use this in practice.
Response:
As a result of this paper, we have only evaluated four aspects of software development of functionality, ease of use, system performance, and security systems, but have not provided examples to illustrate. How to use this system in practice is the focus of the authors' subsequent research.
The article does not offer a deeper comparison of the results with other studies on AI technologies, including GAN-based systems. The discussion of the results is mainly limited to their presentation, without addressing how these findings relate to previous studies or how they may influence future work in this field.
Response:
Articles of lines 83-115 have presented many researches on artificial intelligence technology in applications. They point out that all kinds of software currently on the market are deficient in functionality, ease of use, system performance, and security. The evaluation system of this paper has fully considered the previous results, it expands the functionality of the software and incorporates the evaluation system, so that the enterprises can make the newly developed software to meet the needs of the enterprises based on the evaluation system given by us.
Conclusions: The final conclusions are formulated concisely, but there is a lack of discussion on future research directions and potential improvements in the proposed system.
Response:
This research has constructed a GAN-based evaluation system for AI painting software. On this basis, it provides more reference standards for AI technology companies that are developing related software. The evaluation system established in this study has a potential direction for improvement, i.e., the evaluation system should also reflect whether the software applies the latest technology, including the following aspects:
- frequency of technology updates, including how often the software updates its core algorithms and how often the software version is updated
- the degree of integration of new technologies, i.e., evaluate whether the AI painting software incorporates emerging AI technologies? Such as NLP technology and multimodal generation technology
The article is valuable and contains a solid theoretical foundation and a clear methodology. However, several improvements are recommended to strengthen the clarity of the message and provide a fuller picture of the applied techniques and research outcomes. It is especially important to expand the discussion of the results and to add a broader comparative perspective.
Response:
We thank the reviewers for their suggestions on this paper. The focus of this paper is to construct an evaluation system for AI painting software with generative adversarial network technology and to determine the importance of each index through fuzzy hierarchical analysis and fuzzy comprehensive evaluation to enhance the information of this technology. Further research in this paper, such as the generation of face images as an example, can serve as a basis for the application of this technology to diversified and rich paintings.This can broaden the extension of the discussion of the results and add a wider comparative perspective.
Technical remarks:
Table 2. The caption is incomplete, which may make it difficult for readers to understand the content of the table. It is recommended to complete the description to clearly explain what the table contains and what the data represent.
Response:
Table 2 has been added and re-revised. Scale and description of relative importance in Table 2.
It is worth ensuring that key terms, such as ‘Generative Adversarial Networks (GAN)’ and ‘AI painting software,’ are consistently used throughout the paper. In several places, these terms may be used somewhat variably, which could confuse readers.
Response:
We have ensured that key terms (e.g. Generative Adversarial Networks (GAN) and AI painting software) are used consistently throughout the paper. Please refer to the red word markers in the text.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsI would like to thank the authors for the reponses to the comments. As seen in the responses themselves, the work makes highly theoretical and tehcnical yet connections are not drawn with pedaogical theory and the justification and reasoning of the mathematical decisions are not grounded and justified in the education real (concepts, theories, known best practices)
Author Response
Comments and Suggestions for Authors
I would like to thank the authors for the reponses to the comments. As seen in the responses themselves, the work makes highly theoretical and tehcnical yet connections are not drawn with pedaogical theory and the justification and reasoning of the mathematical decisions are not grounded and justified in the education real (concepts, theories, known best practices)
Response:
This article proposes an evaluation system for solving the 4 difficulties of GNN such as functionality, ease of use, system performance and safety and the importance is ranked by fuzzy hierarchical method. In addition, this paper also focuses on functionality, ease of use, system performance and safety of AI painting software which can increase the competitiveness of the software in the market. Enterprises can save more cost by investing money in the development of software. Whereas, the comments made by the reviewer such as teaching theory, mathematical decision making and reasoning are irrelevant to this thesis so they cannot be answered.
Author Response File: 
Author Response.docx
