Non-Pattern-Based Anomaly Detection in Time-Series
, Anton Kudin 2, Victor R. Kebande 1, Oleksii Baranovskyi 1,2 and Ivan Kudin 2
Round 1
Reviewer 1 Report
The Authors present anomaly detection in time series and conduct extensive experiments.
Shortcomings need to be improved:
1. References look old, making the theme less novel and challenging to attract readers.
2. The authors show the source code at https://github.com/vntkach/anomalydetection. However, the link is broken.
3. The title of Fig. 1-2 seems slightly casual.
4. Lines and axis in some figures are hard to distinguish.
Author Response
The authors appreciate the efforts of the editors and the reviewers in providing valuable comments to enhance the quality of the manuscript. The purpose of this document is to summarize the main changes made to the manuscript and to provide detailed responses to the reviewers’ comments. The reviewers’ comments are placed in boxes, and the corresponding response is positioned below each box. In the revised version, all changes related to the previous submission have been highlighted in blue color. We would like to thank the reviewers for their efforts in helping us improve the quality of this article.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper proposes a non-pattern anomaly detection method that can identify outliers in the time series. It uses a finite state machine approach combined with likelihood probabilities and logistic functions to determine whether anomalies occur in time series data .However, there exist some problems in the paper, which are as follows:
1）The definition of non-pattern anomaly detection in the paper is ambiguous. Especially, the authors should clarify the main difference between normal anomaly detection and non-pattern anomaly detection and state why non-pattern is important to solve?
2）The experimental part should be improved. It is suggested that the authors should conduct fair comparison on different datasets in diversified scenarios with more traditional anomaly detection methods and SOTA methods proposed in recent years.
3）The methodology section is too brief to demonstrate the overall framework and methodology of the paper. While the concrete method is described in part 4 non-pattern anomaly detection (NP-AD). So the sections and structure of the paper should be arranged properly for convenience of reviews and readers.
Author Response
The authors appreciate the efforts of the editors and the reviewers in providing valuable comments to enhance the quality of the manuscript. The purpose of this document is to summarize the main changes made to the manuscript and to provide detailed responses to the reviewers’ comments. The reviewers’ comments are placed in boxes, and the corresponding response is positioned below each box. In the revised version, all changes related to the previous submission have been highlighted in blue color. We would like to thank the reviewers for their efforts in helping us improve the quality of this article.
Author Response File: Author Response.pdf
Reviewer 3 Report
Please find the comments at the attached PDF file.
Comments for author File: Comments.pdf
Author Response
The authors appreciate the efforts of the editors and the reviewers in providing valuable comments to enhance the quality of the manuscript. The purpose of this document is to summarize the main changes made to the manuscript and to provide detailed responses to the reviewers’ comments. The reviewers’ comments are placed in boxes, and the corresponding response is positioned below each box. In the revised version, all changes related to the previous submission have been highlighted in blue color. We would like to thank the reviewers for their efforts in helping us improve the quality of this article.
Author Response File: Author Response.pdf
Reviewer 4 Report
This paper presents non-pattern anomaly detection in time series, which can provide warnings of potential cybersecurity attacks.
Major comments
The authors should explain the advancement of the proposed approach, comparing the existing state-of-the-art approach, to highlight the contribution of this paper.
The author should discuss the metrics such as false-positive rate, F1, ROC-AUC in the text as they are really important in practical scenarios.
Can this approach be viewed as a generalization or extension of change point detection? There is no mention of that.
The code repository link mentioned in the paper could not be located. Verifying the paper's claims is crucial, and running the code is one of the simplest ways to do so. The authors should consider making the code for their method and empirical experiments available to the reviewers and later release it publicly.
Many of the sentences are grammatically strange.
Author Response
The authors appreciate the efforts of the editors and the reviewers in providing valuable comments to enhance the quality of the manuscript. The purpose of this document is to summarize the main changes made to the manuscript and to provide detailed responses to the reviewers’ comments. The reviewers’ comments are placed in boxes, and the corresponding response is positioned below each box. In the revised version, all changes related to the previous submission have been highlighted in blue color. We would like to thank the reviewers for their efforts in helping us improve the quality of this article.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The authors have answered all my questions and i advised the manuscript to be published in the journal.
Author Response
Dear reviewer,
The authors appreciate your time and attention to our paper. Your valuable comment allowed for its improvement. The comment has been addressed as per your suggestion.
Sincerely,
authors
Reviewer 3 Report
Thank you very much for revising the manuscript according to comments.
Author Response
Dear reviewer,
The authors appreciate your time and attention to our paper. This helped us to improve the paper. Thank you for your positive review.
Sincerely,
authors
Reviewer 4 Report
There has been an improvement in the quality of the paper. The content of an author's paper should generally be selected carefully. Despite the authors' elaborate study, their presentation needs to be improved. For instance, Fig 17 is missing. And the aim of figures should be highlighted to the user and followed by a clear conclusion.
Author Response
We take the time to thank the reviewer for this valuable comment. The authors have responded by including the missing Figure 17. The authors have also discussed each of the figures in the improved manuscript.
Author Response File: Author Response.pdf
