|
[交流]
第一篇SCI的修改求助已有5人参与
各位大神好,小弟投了第一篇SCI收到了3个审稿人意见。总体要求大修,但感觉第三个审稿人提了特别多的意见,看得也很细。在这里想请教一下,是否所有的审稿意见都必须接受,另外这是一个快报性质的刊物,最长篇幅不能超过5页,因此审稿人的部分疑问可否只给他解释而不更改文章内容?请各位指教!
以下为审稿意见:
Associate Editor Comments:
Associate Editor
Comments to the Author:
Dear Authors,
One of the reviewers thinks that your work need to be improved in terms of presentation. In particular, you should take care of the connection between the overall sections.
Also the ML approach needs to be explained in more details and in particular you should include answers to the following questions:
1) why the machine learning algorithm provide better solution than the current model.
2) What are the insights of the cross learning and if there is a unified data set that is suitable for all cases, and what are the parameters to consider for testing a complex environment.
Reviewer(s) Comments:
Reviewer: 1
Comments to the Author
This paper studies the problem of evaporation duct height estimation, using an MLP architecture of 5 layers, with RELU activation and ADAM optimizer.
The obtained results are compared to those of a commonly used analytical model, the P-J model, showing supremacy of the proposed approach. Additionally, the proposed network topology is trained with data from one area, and predicts results in another area, thus exploring the correlation between each area's features.
The inversion field of evaporation duct estimation has been studied so far using machine learning techniques, such as SVM and PSO. Hence, it seems that there has been no reference to MLP applied in the evaporation duct diagnoise field. This is the main innovative feature of the present method. The presentation of related work and proposed approach is clear, although there are some misspellings and grammatical errors. For example, the word 'comparasion' should be 'comparison' in page 3.
Based on the mentioned above, I recommend that this manuscript could be accepted after minor revision.
Reviewer: 2
Comments to the Author
Readability of the figures (legends, labels...) should be improved. A comparison to actual values should be included, not only to other theoretical models. At least, infer your precision to real scenarios by finding out the precision of those other theoretical models (in case that those ones have been compared to real world in some references).
Does the climate change influence in this type of prediction approaches?
Reviewer: 3
Comments to the Author
1- First page, left column, line 26 is grammatically not correct
2 First page, left column, line 40, 41 it is not clear the difference between the two approaches and why the author chose the model diagnosis
3- First page, left column, line 50, please try to make a smooth transition between the two paragraphs so as not to have a disconnected flow for the reader
4- First page ,left column, line 50, please comment on why using machine learning instead of the current models is better, is it because of flaws in the model or certain assumptions made ?
5-First page, right column, line 56, please illustrate why as stated in the previous comments, and what are your insights for this ? Also what is the value added compared to the literature for example the approach presented in [9]
6- Second page, left column, line 18, it is recommended to give basic definition of critical gradient, and what variables P and z refer to ? Also if there is a closed form solution, it would be preferable to provide it for a complete understanding of the model.
7- Second page, left column, line 58 , there is a spelling mistake, also the parameters of EDH=f(..) it is not defined
8- Second page, right column, general comment on subsection A, that it is totally unclear, please provide more illustration to this function and why you chose it
9- Second page, right column, line 18, the sentence is incomplete, what is nonlinear ?
10- Second page, right column, subsection B, spelling mistake in the title, also in line 35 and 37
11- Second page, right column, line 55, please explain ADAM algorithm more or put a reference for the readers, and how it is used in the paper is not clear as well.
12- Second page, right column, line 57, “sparse gradient and non stationary targets” please elaborate on the relevance of this to the presented work.
13-Third page, left column, line 26, what is the criteria used to choose this number? Why not 7 layers for example ?
14- Third page, right column, subsection D, spelling mistake in the title, and space missing in line 58
15-Fourth page, left column, line 45, “ environment on NBB is more complex”, please comment about what is the complexity factors?
16- Fourth page, left column, general comment, which database you consider better as training data and more comprehensive to use for all the other areas and why
17- The contribution of the paper includes the use of ADAM, but there is nothing in the results section shows how it improved the performance or the convergence of the optimization algorithm |
|