A Novel Progressive Image Classification Method Based on Hierarchical Convolutional Neural Networks

Round 1

Reviewer 1 Report

The paper has some merits in proposing a progressive image classification method depending on hierarchical convolutional neural networks. However, the paper presentation must be improved. The following issues must be taken care of:

• • The abstract should be made more enlightening, and it should provide a more holistic view of the paper (aim, method, results, contributions, and implications). More technical details should be included. "Results show that HCNNs can obtain better results than classical CNNs and
    13 the existing models based on ensemble learning" How the better is better?
  • The organization/structure of the paper must be considerably improved. Make sure the paper structure is logical and clear to follow. The introduction should emphasize the research motivations. Paper organization should be stated in the last paragraph of the Introduction.
  • The related works should be compared and analyzed to show the research gaps and how the proposed method can overcome the limitation of the conventional approaches. Although authors have classified the related studies into three categories, It is not clear to which category the proposed approach belongs. 
  • The research findings must be better linked back to the existing literature (targeted gap). What are your contributions and how does the proposed method outperforms the conventional? The of presenting the results in tables is not sufficient. It is better to use graphs to improve the readability of the results and prove the effectiveness of the proposed approach.
  • The interpretation and discussion of the research findings are not well developed. It had been nice if the research could end up with some clear propositions. The theoretical and practical implications should be further elaborated on. Is some further research needed? How should researchers act based on your results?
  • The use of abbreviations should be carefully checked. Make sure that all abbreviations are explained the first time they are used and that introduced abbreviations really are used throughout the paper. The symbols should be defined after each equation and a table of notations is required.
  • The readability of the paper must be considerably improved. This issue concerns both grammar, sentence structure, as well as the overall story. Professional proofreading services should be considered.
  • The reference system should be carefully checked. Make sure that citations are made in a uniform and correct fashion throughout the paper and avoid the citation of unrelated references and self-citation.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Title: A Novel Progressive Image Classification Method Based on Hierarchical Convolutional Neural Networks.

The current paper covered an interesting topic to explore Hierarchical Convolutional Neural Networks (HCNNs) for image classification. The authors have analyzed the HCNN and CNN while trying to proof HCCN are better than CNN. However, the following queries and suggestion are necessary to improve the quality of manuscript:

• There are some typo and grammar mistakes which need to be correct. The structure and graphical presentation of results is very good.
• Improve the abstract with key results. Literature should according to timeline. Please report new studies, if possible.
• In introduction section, add some other materials related to the deep learning for medical diagnosis with convolutional neural networks with proper referencing. For example: https://link.springer.com/article/10.1007/s00330-019-06344-z ,  https://doi.org/10.3390/cancers11091235 , https://link.springer.com/article/10.1007/s00432-019-03098-5. https://link.springer.com/chapter/10.1007/978-981-15-6321-8_11, https://www.sciencedirect.com/science/article/pii/B9780128245361000393, https://link.springer.com/chapter/10.1007/978-981-15-6321-8_14 , https://www.researchgate.net/profile/Rasool-Fakoor/publication/281857285_Using_deep_learning_to_enhance_cancer_diagnosis_and_classification/links/5982f029458515a60df82098/Using-deep-learning-to-enhance-cancer-diagnosis-and-classification.pdf
• Why are you choosing HCNN instead of CNN? What is the significance of this? Provide the reference for this.
• Figure 1 has a very poor quality and explanation. Improve this figure.
• What is the reason of choosing HCNN and why it's better ?
• Check the captions of tables and figure? There are some typo mistakes.
• Improve the discussion deeply with scientific point of view.
• Conclusion should be critical based on the results.

Overall, the quality of the manuscript is good and can be accepted for publication after this minor revision.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Thanks to the authors for their efforts to revise the manuscript.

The authors have addressed my concerns and the revised version has been improved. An overall check is required especially for the newly added paragraphs. 

Author Response

1. The authors have addressed my concerns and the revised version has been improved. An overall check is required especially for the newly added paragraphs.  
Response: Thank you very much for this comment, we have carefully check out the paper, and tried our best to make corrections in the revision (Noted in blue color).