3D Model Generation on Architectural Plan and Section Training through Machine Learning
Round 1
Reviewer 1 Report
The article describes the case of using the Generative Adversarial Network (GAN), in which the author uses the algorithm developed by NVIDIA in 2018 called StyleGAN. After reading the text, I pay attention to the following flaws:
There are no references to literature items in the text. References at the end are incorrectly formatted. Many of them have no dates. The article does not have the correct structure: Introduction, Materials and Methods, Results, Discussion, Conclusions. StyleGAN has been described in detail in the article [1]. The author's contribution consisted only of using other types of images to train the network. It is not clear what the author wanted to achieve. The choice of architectural images is not properly justified. It even seems that he was accidental. The results obtained do not bring anything significant to the current state of knowledge about GAN. The positive feature of the presented article is the popularization of innovative machine learning techniques, which are undoubtedly GAN. However, there are serious doubts as to whether such article can be described as scientific. It is rather a popular science text.Author Response
Please see the attachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper proposes an application of StyleGAN to learn and simulate the relationship between two architectural plans or sections and generate serialized transformation pictures to build a 3d model. Results show that the 3d model generation is difficult to be directly used in 3d space modelling and that they can inspire new design methods and more potential possibility of architectural plan and section design.
As a general comment, I think the paper makes an interesting contribution to the literature by providing this application. The results section is clearly explained. At the same time, I think the paper needs some improvements before to be published in this journal.
Broad comments
The abstract should be improved adding the main findings of the paper. An introduction section should be added, where the state of the art in the field should be cited and the research questions should be defined. The different methodologies used in literature should be cited. The following papers makes an interesting comparison among different methods (traditional, neural networks and support vector machine) applied in different fields. I would suggest the author to add at least a paragraph in the introduction where these papers should be mentioned. Rafał Weron,“Electricity price forecasting: A review of the state-of-the-art with a look into the future”, International Journal of Forecasting, Volume 30, Issue 4, Pages 1030–1081, 2014 Silvano Cincotti, Giulia Gallo, Linda Ponta, Marco Raberto, “Modelling and forecasting of electricity spot-prices: Computational intelligence vs classical econometrics”, AI Communications, Volume 27, Issue 3, Pages 301-314, 2014 A section where the author describes the methodology used should be added before the different experiments. The author should add a section where the data used in the experiments are described. The author should explain how the percentage of the partition of the data in training and validation set are chosen. Do the results change if these percentages change?Specific comments
Before using the acronym GAN, you have to specify its meaning. In section 1, GAN (Generative Adversarial Network) is used without defining the meaning.Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
After reviewing the corrections applied by the author, I conclude that the article is suitable for publication.
Author Response
Thank you for reviewing.
Reviewer 2 Report
The paper has been improved following the referees’ suggestions.
I would like to ask the authors why they did not include the second reference suggested, i.e. Silvano Cincotti, Giulia Gallo, Linda Ponta, Marco Raberto, “Modelling and forecasting of electricity spot-prices: Computational intelligence vs classical econometrics”, AI Communications, Volume 27, Issue 3, Pages 301-314, 2014.
According to me this reference should be added before to publish the paper.
Author Response
The second reference has been added. Last round I haven't finish this article so I'm not so sure if I should cite it.
