Optimization of the Random Forest Hyperparameters for Power Industrial Control Systems Intrusion Detection Using an Improved Grid Search Algorithm

Round 1

Reviewer 1 Report

The authors propose a hyperparameter optimization method based on an improved grid search algorithm for improving the detection performance of the Random Forest algorithm. Several parameters and values are assessed and the best set of parameters was chosen with significant gains in speed in comparison to the traditional grid search. Besides, several intrusion detection metrics were maximized with the proposed algorithm.

As a suggestion for the authors, they could compare (or at less discuss) the potential of using metaheuristics such as Tabu Search and other similar ones to replace the Grid Search algorithm and optimizing still more the computational time.

Minor comments: please review all text for grammar and text issues. Some examples are listed below.

- Line 12: "information system" > "information systems"
- Line 27: "smart grid" > "smart grids"
- Line 82: "a RF model" > "an RF model" (review all text, such as line 92)
- Line 84: "each single" > "every single"
- Line 100: "full text" > "full-text"
- Line 221: "of grid" > "of the grid..."

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Title: A Novel Method of Hyperparameter Optimization-Based Random Forest Algorithm

for Power Industrial Control Systems Intrusion Detection.

Authors: Ningyuan Zhu * , Chaoyang Zhu , Liang Zhou , Yayun Zhu , Xiaojuan Zhang

 

1- the title is not well written.

proposed title: Optimization of the Random Forest hyperparameters for Intrusion Detection using an Improved Grid Search Algorithm.

2- last sentence in abstract is long and not well written: use a short sentence with the correct punctuation.

3-use acronym of improved grid search algorithm (IGSA)

4- add key word: hyperparameter optimization

5-explain the equation parameters.

6- explain the use of AUC metric.

7-explain why you use many evaluation metrics (accuracy, F1-score, recall, precision).

8-add other references related to the results and discussion.

9- use k-fold cross validation (k=5 or k=10) to evaluate your results.

10- explain the training/testing splitting of your dataset.

11-is the dataset is balanced or not?

12- the conclusion is short.

13- explain how the proposed approach boosts the speed of calculation

 from O(nm) to O(n×m) in the conclusion

14- add other references to RF method.

15- the form of the references must be respected

16- add other references related to the optimization of the parameters Random Forest.

17-reference 27 to 34 must be deleted.

18- plagiarism: 7% to 10 % (can be improved).

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

The majority of the comments are implemented and the quality of the paper is improved.