A Novel Approach to Classify Telescopic Sensors Data Using Bidirectional-Gated Recurrent Neural Networks

Round 1

Reviewer 1 Report

The article is written very carefully and interestingly. The article is aimed at a wide range of readers because it is written in an exquisite style. The authors propose a deep learning approach to separate red giant branches from red clump stars using asteroseismic data. The authors come from the computer science community, and cross-disciplinary investigations can lead to fruitful and novel results.

I accept the paper with some minor concerns that need to be addressed:

1.      Is the dataset created by authors, or existing dataset is used to conduct research experiments? Please clarify it.

2.      Why the 80:20 data split ratio is used for deep learning model building?

3.      Round the log loss score in tables 8 and 9 to three digits, like 10.6.

- Extend the related work section with some of the recent published and related papers.

- Improve the quality of figure 8

Make the conclusion more comprehensive to the readers

-

Author Response

Attached herewith for your reference.

Author Response File: Author Response.docx

Reviewer 2 Report

Accept

Author Response

Attached herewith for your reference.

Author Response File: Author Response.docx

Reviewer 3 Report

The main aim of this research study is to classify the RGB and HeB 25 in Asteroseismology using a deep learning approach. A novel Bidirectional-Gated recurrent units 26 and Recurrent neural networks (BiGR) based deep learning approach is proposed. The proposed 27 model achieved a 93% accuracy score for Asteroseismology classification. The proposed technique 28 outperforms other state-of-the-art studies. The analyzed fundamental properties of RGB and HeB 29 are based on the frequency separation of modes in consecutive order with the same degree, maxi- 30 mum oscillation power frequency, and mode location. The Asteroseismology Exploratory Data 31 Analysis (AEDA) is applied to find critical fundamental parameters and patterns that accurately 32 infer from the asteroseismology dataset.

I recommend a minor revision. The study is worthy for publication in Applied Sciences subject to the following comments:

1. Article needs a thorough recheck for grammar and typological errors

2. Please elaborate on key findings in the abstract. Please remove the old and unnecessary references and add the latest references.

3. Conclusion section looks poor.

Author Response

Attached herewith for your reference.

Author Response File: Author Response.docx

Reviewer 4 Report

The manuscript presents a study of the RGB and HeB classification in Asteroseismology that uses deep learning based on a novel BiGR approach. After reading whole the manuscript, I thought that it is a fascinating approach, and will be widely interested in the Astronomical science community. The paper is well-structured, logical, and easy to understand. Therefore, I recommend the paper to publish in Applied Science. However, there are some issues in the manuscript, which I think should be addressed properly before further consideration. My detailed comments are listed below.

 

1.     The data used in machine learning should describe in more detail. It is better to briefly introduce the input data of the study, giving the potential reader some graphs to illustrate.

2.     Did the authors use other techniques proposed in Table 12 (e.g., CNN, ResNet-UNet) tested on the dataset used in this study? If it does, what are the comparative results?

 

3.     The authors need to point out the current problems that exist when astronomers use traditional methods to classify star oscillations and explain why the advantages of the method proposed in this paper can help astronomers better analyze data. Is the data classified by the authors, and is it open source? Open source can let more astronomers use it.

Author Response

Attached herewith for your reference.

Author Response File: Author Response.docx