SCORN: Sinter Composition Optimization with Regressive Convolutional Neural Network

Round 1

Reviewer 1 Report

In this paper the authors proposed representation learning capability of the deep networks to optimize sinter compositions from the sinter production. It was suggested a sinter composition optimization model based on an RCNN. Based on the experiments performed by the authors, it was found that the proposed approach is able to predict the sinter composition changes with a higher R² value.

This paper is a well-written and well-structured. It deals with an important problem regarding optimization of sinter composition. Overall, the manuscript is well written, the objectives clearly stated, the introduction is relevant, and theory based. Sufficient information about the previous study findings is presented for readers to follow the present study rationale and procedures. Theoretical and experimental methods are advanced, data statistically analyzed, the conclusions well supported by the data presented. Therefore, in my opinion this paper is suitable for publication in the present form. 

 

 

Author Response

Thanks for the valuable comments from Reviewer 1. We appreciate your dedicated time and effort in providing feedback on our manuscript. We also re-checked and improved the language in our revised manuscript. All typos and minor grammar errors are carefully addressed.

Reviewer 2 Report

In this study, the authors proposed convolutional neural network architecture for Sinter Composition Optimization. It is a necessary problem and the authors could reach a promising performance on the model. However, some major points should be addressed as follows:

1. The study was conducted on only one dataset without any external validation. Thus, the authors should add some external validation data to show the efficiency of model on different data.

2. How did the authors conduct hyperparameter tuning of the models?

3. The authors are suggested to conduct cross-validation on the training process.

4. Uncertainties of models should be reported.

5. Fig. 3 should be included with the validation curve together.

6. The performance looked inconsistent and chaotic (Fig. 4). How to solve this problem?

7. CNN is well-known and has been used in previous studies i.e., PMID: 34915158, PMID: 31380767. Thus, the authors are suggested to refer to more works in this description to attract a broader readership.

8. Source codes should be provided for replicating the study.

9. Running time of models should be reported.

10. English language should be re-checked and improved.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

My previous comments have been addressed.