Boosting-based Models with Tree-structured Parzen Estimator Optimization to Detect Intrusion Attacks on Smart Grid

Khoei, Tala Talaei; Ismail, Shereen; Kaabouch, Naima

doi:10.1109/uemcon53757.2021.9666607

Cited by 13 publications

(10 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The bridge dimensions establish that the IE signals were collected from a large surface area, demonstrating that the IE signals in this study were generalized and diverse compared to previous research [24]. A balanced dataset can be crucial for developing robust machine learning and artificial neural networks [30][31][32][33][34][35]. We used four training models based on IE signals in two classes.…”

Section: Impact-echo Dataset and Classificationmentioning

confidence: 78%

“…SDNET2021 contains 1573 IE signals from four bridges in total. Under sampling is an approach used to create a balanced dataset by maintaining all of the data in the smaller class and reducing the size of the data in the larger class [30][31][32][33][34][35]. A balanced training set was created using an under-sampling strategy in this study.…”

Section: Impact-echo Dataset and Classificationmentioning

confidence: 99%

See 1 more Smart Citation

Comparison between Supervised and Unsupervised Learning for Autonomous Delamination Detection Using Impact Echo

Jafari

Dorafshan

2022

Remote Sensing

View full text Add to dashboard Cite

Impact echo (IE) is a non-destructive evaluation method commonly used to detect subsurface delamination in reinforced concrete bridge decks. Existing analysis methods are based on frequency domain which can lead to inaccurate assessments of reinforced concrete bridge decks since they do not consider features of the IE signals in the time domain. The authors propose a new method for IE classification by combining features in the time and the frequency domains. The features used in this study included normalized peak values, energy, power, time of peaks, and signal lengths that were extracted from IE signals after they are preprocessed. We used a dataset containing IE data collected from four in-service bridges, annotated using chain dragging. A support vector machine (SVM) classifier was constructed using combined features to classify IE signals. A 1DCNN with unfiltered IE signals and a two-dimensional CNN using wavelet scalograms (2D representations of unfiltered IE signals) were also used to classify IE signals. The SVM model performed significantly better than the other models, with an accuracy rate, true positive rate, and true negative rate of 97%, 92%, and 98%, respectively. The SVM model also generated more accurate defect maps for all investigated bridges. IE data from the Federal Highway Administration’s InfoBridge website were used to investigate the efficacy of the developed models. The investigation yielded promising results for the proposed SVM model when used for a new set of IE data.

show abstract

Section: Impact-echo Dataset and Classificationmentioning

confidence: 78%

Section: Impact-echo Dataset and Classificationmentioning

confidence: 99%

Comparison between Supervised and Unsupervised Learning for Autonomous Delamination Detection Using Impact Echo

Jafari

Dorafshan

2022

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…The results show that the stacking technique provides satisfactory results. In [114], the authors use boosting ensemble technique along with Tree-structured Parzen Estimator Optimization to detect and classify attacks on smart grid. The authors show that optimization techniques improve the detection performance of the mode.…”

Section: Ai-based Techniquesmentioning

confidence: 99%

Cyber-Security of Smart Grids: Attacks, Detection, Countermeasure Techniques, and Future Directions

Khoei¹,

Slimane²,

Kaabouch³

2022

Self Cite

View full text Add to dashboard Cite

One of the significant challenges that smart grid networks face is cyber-security. Several studies have been conducted to highlight those security challenges. However, the majority of these surveys classify attacks based on the security requirements, confidentiality, integrity, and availability, without taking into consideration the accountability requirement. In this survey paper, we provide a classification of attacks based on the OSI model and discuss in more detail the cyber-attacks that can target the different layers of smart grid networks communication. We also propose new classifications for the detection and countermeasure techniques and describe existing techniques under each category. Finally, we discuss challenges and future research directions.

show abstract

“…It should be noted that most of the proposed methods have only included CIC-IDS2017 or CSE-CIC-IDS2018 in their analysis. Reference [13] included CIC-DoS2017 (besides IDS2017 and IDS2018), references [14] and [15] were validated on CIC-DDoS2019 and reference [16] included all four CIC datasets.…”

Section: B Proposed Ml-ids Detection and Feature Selection Systemsmentioning

confidence: 99%

“…CIC Collection (IDS17, IDS18, DoS17, DDoS19) [27] 2020 PSH flag count, fwd packet length min, bwd packet length min, down/up ratio [28] 2020 protocol, down/up ratio, active std [13] 2021 PSH flag count, active std [16] 2021 fwd packet length min, packet length min, PSH flag count, ACK flag count [14] 2021 protocol, bwd packet length min [15] 2021 packet length min, bwd packet length min [29] 2022 PSH flag count, packet length min, protocol, fwd packet length min, bwd packet length min active std, idle std, ACK flag count, ECE flag count, RST flag count, down/up ratio [33]) proposed for cyber security ML systems on six criteria. They conclude that white-box techniques (full model access) such as integrated gradients and layer-wise relevance propagation comply best with the tested criteria.…”

Section: Yearmentioning

confidence: 99%

Discovering Non-Metadata Contaminant Features in Intrusion Detection Datasets

D’hooge

Verkerken

Wauters

et al. 2022

2022 19th Annual International Conference on Privacy, Security &Amp; Trust (PST)

View full text Add to dashboard Cite

Most newly proposed detection methods in intrusion detection incorporate machine learning models to distinguish between benign and malicious traffic. The models are validated on a handful of academic datasets and ranked based on their classification performance. This article aims to demonstrate that unbeknownst to the new models' authors, there are features in these datasets which heavily bias the results and obscure a realistic, reliable estimate of the separability of the datasets. This paper proposes a methodology to estimate the contaminating influence of a dataset's features based on the concept of blind generalization. The novel methodology is subsequently used to assess the features of six widely adopted intrusion detection datasets. In each dataset, several features show a pattern where regardless of training attack class, the models blindly generalize towards all available attack classes with nearly identical classification metrics. These features provide undeserved boosts in the baseline classification scores for each dataset. By themselves, some contaminant features even push these baselines upwards of 90% accuracy (balanced).

show abstract

Boosting-based Models with Tree-structured Parzen Estimator Optimization to Detect Intrusion Attacks on Smart Grid

Cited by 13 publications

References 15 publications

Comparison between Supervised and Unsupervised Learning for Autonomous Delamination Detection Using Impact Echo

Comparison between Supervised and Unsupervised Learning for Autonomous Delamination Detection Using Impact Echo

Cyber-Security of Smart Grids: Attacks, Detection, Countermeasure Techniques, and Future Directions

Discovering Non-Metadata Contaminant Features in Intrusion Detection Datasets

Contact Info

Product

Resources

About