Fuzzy Classifier Based on Mamdani Inference and Statistical Features of the Target Population
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsSummary: The study introduces a fuzzy classifier that uses a Mamdani inference system and chaotic maps (logistic, Bernoulli, tent) to model random features of a population. These maps generate statistical inputs for the classifier, which successfully groups objects based on their variability. Experimental results confirm the system’s ability to adapt and accurately classify by adjusting membership functions.
Comments: My comments are mentioned below:
- Revise the abstract with the research motivation and new outcomes.
- There are few articles that are similar to the present research. The authors must provide the clarification about the new ideas of this research. How the manuscript differs from the following articles:
- Weihong, Z., Shunqing, X. and Ting, M., 2010, November. A fuzzy classifier based on Mamdani fuzzy logic system and genetic algorithm. In 2010 IEEE Youth Conference on Information, Computing and Telecommunications (pp. 198-201). IEEE.
- Gayathri, B.M. and Sumathi, C.P., 2015, December. Mamdani fuzzy inference system for breast cancer risk detection. In 2015 IEEE international conference on computational intelligence and computing research (ICCIC) (pp. 1-6). IEEE.
- Lucchese, L.V., de Oliveira, G.G. and Pedrollo, O.C., 2021. Mamdani fuzzy inference systems and artificial neural networks for landslide susceptibility mapping. Natural Hazards, 106(3), pp.2381-2405.
It is astonished to see that the authors have not cited the articles. I suggest the authors to cite them mentioning how their manuscript different from them.
- Figure 4-10: I suggest enhancing the size of the figures so that they are visible clearly to the readers.
- There are several punctuation errors. Equations should be ended with a comma or full stop. Check all the grammatical errors to fix them.
- Results of section 4 should be reflected in Discussion section. Please rewrite Discussion section.
- Mention some application, in conclusion part, for real world situation with a future direction of the present work.
Author Response
Comments of Reviewer #1:
Summary: The study introduces a fuzzy classifier that uses a Mamdani inference system and chaotic maps (logistic, Bernoulli, tent) to model random features of a population. These maps generate statistical inputs for the classifier, which successfully groups objects based on their variability. Experimental results confirm the system’s ability to adapt and accurately classify by adjusting membership functions.
Comments: My comments are mentioned below:
RQ#1. Revise the abstract with the research motivation and new outcomes.
Answer:
We appreciate the reviewer's comment.
Action:
We have revised the abstract again to reflect corrections and supplemented it with a paragraph describing the motivation for the research and to describe the results obtained. We have also added a few lines mentioning some applications that were added to the discussion of our work.
See: PAGE 1 OF THE MANUSCRIPT.
RQ#2. There are few articles that are similar to the present research. The authors must provide the clarification about the new ideas of this research. How the manuscript differs from the following articles:
Weihong, Z., Shunqing, X. and Ting, M., 2010, November. A fuzzy classifier based on Mamdani fuzzy logic system and genetic algorithm. In 2010 IEEE Youth Conference on Information, Computing and Telecommunications (pp. 198-201). IEEE.
Gayathri, B.M. and Sumathi, C.P., 2015, December. Mamdani fuzzy inference system for breast cancer risk detection. In 2015 IEEE international conference on computational intelligence and computing research (ICCIC) (pp. 1-6). IEEE.
Lucchese, L.V., de Oliveira, G.G. and Pedrollo, O.C., 2021. Mamdani fuzzy inference systems and artificial neural networks for landslide susceptibility mapping. Natural Hazards, 106(3), pp.2381-2405.
It is astonished to see that the authors have not cited the articles. I suggest the authors to cite them mentioning how their manuscript different from them.
Answer:
We appreciate the reviewer's comment and agree that additional references should be added to highlight the value of our work.
Action:
The suggested articles were reviewed in detail to assess their relevance to our work. Some similarities were found in relation to the use of statistics as part of the design or evaluation of the results in the presented work.
We added a paragraph considering the suggested articles to improve our manuscript. We used these articles and additional findings from other manuscripts to support the novelty of our work with three well-defined points.
See: PAGE 3 OF THE MANUSCRIPT, SECTION 1.
RQ#3. Figure 4-10: I suggest enhancing the size of the figures so that they are visible clearly to the readers.
Answer:
We appreciate the reviewer's comments and agree that the figures should be made larger for easier observation by readers.
Action:
We have increased the size of Figures 3 to 10 by 75%. We have also moved them closer to the paragraph in which they are mentioned, as necessary.
See: PAGES 10, 12, 14, 15, 16, and 17 OF THE MANUSCRIPT, SECTION 3 AND 4.
RQ#4. There are several punctuation errors. Equations should be ended with a comma or full stop. Check all the grammatical errors to fix them.
Answer:
We appreciate the reviewer's comment and apologize for punctuation errors in the equations that need to be corrected.
Action:
We have carefully reviewed the author’s guide in the journal to confirm the rules regarding this matter. We have revised not only the equations, but the entire document, correcting all grammatical errors.
See: PAGE 5 OF THE MANUSCRIPT, SECTION 2.
RQ#5. Results of section 4 should be reflected in Discussion section. Please rewrite Discussion section.
Answer:
We appreciate the reviewer's comment.
We had not noticed the deficiencies in writing and consistency in our manuscript that the reviewer pointed out.
Action:
We have organized the Discussion Section into two subsections. Section 5.1 describes the interpretation of the experimental results, while Section 5.2 presents a comparison with related work.
See: PAGES 17 AND 18 OF THE MANUSCRIPT, SECTION 5.
RQ#6. Mention some application, in conclusion part, for real world situation with a future direction of the present work.
Answer:
We appreciate the reviewer's comment and agree that adding real applications is necessary to determine the direction of future work.
Action:
We have added a paragraph at the end of the Conclusions Section. In it, we describe two possible applications closely related to the focus of our work. This provides a clearer picture of possible future work and its real-world applications.
See: PAGES 20 and 21 OF THE MANUSCRIPT, SECTION 6.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsThe research addresses innovative topics in data mining and computational intelligence. However, the authors should more clearly highlight the performance benefits of the proposed fuzzy-based classification algorithm in the introductory section. For example, many authors have used the fuzzy-cmeans fuzzy clustering algorithm as a classifier, mapping the resulting clusters and determining the optimal number of clusters using validity indices. Fuzzy-cmeans has the advantage of being computationally fast. What limitations and performance pain points of fuzzy-cmeans are addressed and overcome by the proposed classification algorithm?
I recommend reviewing the formulas to avoid ambiguity and to make the manuscript easier to read. It is necessary to declare each variable and index (for example, µ and x in equation (1), the index n in equation (1), and the parameter µ in equation (4)). Furthermore, different variables and parameters must have different symbols (therefore, µ in equation (4) must be reported with a different symbol, to avoid confusion with the parameter µ in equation (1)).
Section 3 lacks an initial figure showing the architectural scheme of the proposed method, which shows the functional components and the relationships between them. It is not very meaningful to describe the well-known Mamdani FIS in Section 3, as well as its differences from the well-known Tagaki-Sugeno FIS. Instead, it is appropriate to describe the processes that perform the algorithm's functional components and how they are related.
What is the computational complexity of Algorithm 1? Furthermore, the authors must explain why the algorithm's for loop is iterated 100,000 times. What motivates the choice of an end-of-loop parameter of 1x10^5, and how does this choice affect the algorithm's computational complexity?
Comparative tests with other fuzzy-based classification algorithms are lacking. The authors need to conduct these comparisons to determine the method's performance benefits in terms of classification accuracy and precision.
Author Response
Comments of Reviewer #2:
RQ#1. The research addresses innovative topics in data mining and computational intelligence. However, the authors should more clearly highlight the performance benefits of the proposed fuzzy-based classification algorithm in the introductory section. For example, many authors have used the fuzzy-cmeans fuzzy clustering algorithm as a classifier, mapping the resulting clusters and determining the optimal number of clusters using validity indices. Fuzzy-cmeans has the advantage of being computationally fast. What limitations and performance pain points of fuzzy-cmeans are addressed and overcome by the proposed classification algorithm?
Answer:
We appreciate the reviewer's comment and agree on the need to add a comparison with the fuzzy c-means algorithm.
Action:
We have added a paragraph to the Introduction Section that mentions the advantages and disadvantages of the fuzzy c-means algorithm and compares it with the proposed classification algorithm. This reinforces the contributions of our article in comparison with works that have used fuzzy c-means.
See: PAGES 2 and 3 OF THE MANUSCRIPT, SECTION 1.
RQ#2. I recommend reviewing the formulas to avoid ambiguity and to make the manuscript easier to read. It is necessary to declare each variable and index (for example, µ and x in equation (1), the index n in equation (1), and the parameter µ in equation (4)). Furthermore, different variables and parameters must have different symbols (therefore, µ in equation (4) must be reported with a different symbol, to avoid confusion with the parameter µ in equation (1)).
Answer:
We appreciate the reviewer's comment and agree that the formulas need improvement to avoid ambiguity.
Action:
We have corrected each of the formulas corresponding to the chaotic maps used in our work. To make them easier to read, we used different variables in each one and we added a brief definition for each one.
See: PAGE 5 OF THE MANUSCRIPT, SECTION 2.
RQ#3. Section 3 lacks an initial figure showing the architectural scheme of the proposed method, which shows the functional components and the relationships between them. It is not very meaningful to describe the well-known Mamdani FIS in Section 3, as well as its differences from the well-known Tagaki-Sugeno FIS. Instead, it is appropriate to describe the processes that perform the algorithm's functional components and how they are related.
Answer:
We appreciate the reviewer's comment and agree that Section 4 should include an introductory figure that outlines the proposed method.
Action:
We have added a block diagram at the beginning of this section showing the architecture of the proposed method and considers the included algorithm. We also consider it necessary to include the basic information about Mamdani and Tagaki-Sugeno ISCs. We have also added a brief description of how the components of the algorithm relate to each other in the process next to the outline.
See: PAGE 8 OF THE MANUSCRIPT, SECTION 3.
RQ#4. What is the computational complexity of Algorithm 1? Furthermore, the authors must explain why the algorithm's for loop is iterated 100,000 times. What motivates the choice of an end-of-loop parameter of 1x10^5, and how does this choice affect the algorithm's computational complexity?
Answer:
We appreciate the reviewer pointing out the need to justify the complexity of the algorithm and how it is applied within the proposed system. We had overlooked this improvement to our manuscript.
Action:
We have added two paragraphs explaining the reasons why the algorithm is iterated 100,000 times, and why we chose the indicated value for the end of the loop.
See: PAGES 11 and 12 OF THE MANUSCRIPT, SUBSECTION 3.4.
RQ#5. Comparative tests with other fuzzy-based classification algorithms are lacking. The authors need to conduct these comparisons to determine the method's performance benefits in terms of classification accuracy and precision.
Answer:
We appreciate the reviewer's comment.
Although we believe that a comparison with other algorithms is necessary to evaluate the performance benefits of our proposal, experiments for evaluation were not considered within the scope of this work due to the complexity of the process. Additionally, we could not find any parameters that could contribute more accurate data to a comparison.
Action:
We have dedicated a paragraph near the end of the discussion section to justify the decision not to conduct experiments comparing the algorithms. Similarly, we considered it more appropriate to compare the methods used by other authors when employing statistics in their designs.
See: PAGES 18 and 19 OF THE MANUSCRIPT, SUBSECTION 5.2.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe revised version is okay as the authors have carefullay answered my queries.
The revised manuscript is acceptible.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have taken into account all my suggestions, improving the quality of their manuscript. I consider this paper publishable in its current version
