Domain Generalization Model of Deep Convolutional Networks Based on SAND-Mask
Round 1
Reviewer 1 Report
This paper proposed an improved DCNG model (deep convolutional network generalization) for SAND-Mask (Smoothed AND (SAND)-masking) by analyzing the characteristics of data in the field of fault diagnosis; the gradient variance of each sample is replaced by the total gradient variance of all samples in a batch to calculate the parameter ?, to calculate the gradient variance. Three experiments were performed on three data sets to verify the validity of the DCNG model. The experimental results show a modification in this problem compared with other models; however, the article should be revised as follows:
- English writing needs consideration and can be improved by a native.
- The abstract can be re-written, and principal research gaps and contributions are unclear.
- Although the introduction is well-organized, the existing research gaps were not adequately discussed and listed in the introduction section.
- The introduction section would be great to include this work's main contributions and novelty. Please list them one by one.
- The formulas need to be numbered in the whole paper.
- In the Tables, the best-found results can be bold.
- In the end, adding a conclusion section can be helpful.
- Testing the performance of the proposed model in more extensive (larger) datasets is recommended.
- Please list and discuss the applied hyper-parameters of the proposed model.
- Considering some relevant references for hyper=parameters tuning methods can be beneficial, such as a) An evolutionary deep learning method for short-term wind speed prediction: A case study of the lillgrund offshore wind farm. arXiv preprint arXiv:2002.09106. b) A method for state-of-charge estimation of lithium-ion batteries based on PSO-LSTM. Energy, 234, 121236. c) Prediction of the NOx emissions from thermal power plant using long-short term memory neural network. Energy, 192, 116597.
Author Response
Please see the attachment
Author Response File: 
Author Response.docx
Reviewer 2 Report
The English is so bad that is too much of a struggle to read this article. Authors should submit something readable if they want a review.
Author Response
Point 1: The English is so bad that is too much of a struggle to read this article. Authors should submit something readable if they want a review.
Response 1: The content in the paper has been revised.
Reviewer 3 Report
The abstract needs to be revised. No quantitative information is present. Also, it is lacking the criteria of the abstract. For example, it should shortly present the background, objective, methodology, results, and conclusions.
The introduction should be concise and focused. Also, some of the most relevant literature needs to be added. As an example, the authors are suggested to take help from this article; "Determining the factors affecting the boiling heat transfer coefficient of sintered coated porous surface". This study presents artificial intelligence-related information.
The aims and motivation of the study should be clear.
The authors are suggested to carefully check the manuscript for grammatical errors.
The conclusion section needs to be revised. For instance, it should present the most important findings of the study.
Author Response
Please see the attachment
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The highlighted version of the revision should be submitted.
Author Response
Point 1: The highlighted version of the revision should be submitted.
Response 1: According to your suggestions, I have highlighted the revised part of the paper.
Reviewer 2 Report
Authors have dealed with most comments in the previous review. The discussion section is still too limited, a more detailed discussion would help the quality of the paper. Also some terms are not clear, what does "complex perfomance" mean? ( However, in scenarios where multiple conditions cause working conditions to change, the performance is more complex and more difficult to be generalized.)
Article can be published after a minor revision.
Author Response
Point 1: Authors have dealed with most comments in the previous review. The discussion section is still too limited, a more detailed discussion would help the quality of the paper. Also some terms are not clear, what does "complex perfomance" mean? ( However, in scenarios where multiple conditions cause working conditions to change, the performance is more complex and more difficult to be generalized.)
Response 1: I have revised and highlighted the discussion section.
Reviewer 3 Report
I suggested you to update the literature review part and improve the methodology especially in terms of graphical representation. Please see my comments carefully.
Author Response
Point 1: I suggested you to update the literature review part and improve the methodology especially in terms of graphical representation. Please see my comments carefully.
Response 1: According to your suggestions, I have modified the paper and highlighted it.
