Unsupervised Stacked Autoencoders for Anomaly Detection on Smart Cyber-physical Grids

Al-Abassi, Abdulrahman; Sakhnini, Jacob; Karimipour, Hadis

doi:10.1109/smc42975.2020.9283064

Cited by 26 publications

(8 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For such events, a whole data set training should be performed, followed by the use of realtime data events to validate and test the algorithm's efficacy. By shifting from an offline to an online environment, DL techniques such as autoencoder-based neural networks [186], generative adversarial networks (GAN) [187], and one-class support vector machines (OCSVMs) [188] can forecast the occurrence and type of fault. Moreover, a lot of the applications listed in [189] essentially evaluated the DL techniques used in electrical PS as a whole, but they may also be used successfully to investigate PS transmission as well as the side of distribution where PMUs and µPMUs are installed.…”

Section: Discussion and Future Trendsmentioning

confidence: 99%

A Review of Machine Learning Approaches in Synchrophasor Technology

Lal

Ramesh

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

Section: Discussion and Future Trendsmentioning

confidence: 99%

A Review of Machine Learning Approaches in Synchrophasor Technology

Lal

Ramesh

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…There is no consensus, and each NN represents different aspects of the surveyed data. In contrast, Al-Abassi et al [ 36 ] split training data in two balanced sets that are used to train two AEs. Those AEs are used for feature reduction, and the hidden layer’s output is used to train subsequent random forest (RF) classifiers.…”

Section: Related Workmentioning

confidence: 99%

Autoencoder-Ensemble-Based Unsupervised Selection of Production-Relevant Variables for Context-Aware Fault Diagnosis

Kaupp

Humm

Nazemi

et al. 2022

Sensors

View full text Add to dashboard Cite

Smart factories are complex; with the increased complexity of employed cyber-physical systems, the complexity evolves further. Cyber-physical systems produce high amounts of data that are hard to capture and challenging to analyze. Real-time recording of all data is not possible due to limited network capabilities. Limited network capabilities are the reason for a chain of faults introduced via active surveillance during fault diagnosis. These introduced faults may slow down production or lead to an outage of the production line. Here, we present a novel approach to automatically select production-relevant shop floor parameters to decrease the number of surveyed variables and, at the same time, maintain quality in fault diagnosis without overloading the network. We were able to achieve higher throughput, mitigate communication losses and prevent the disruption of factory instructions. Our approach uses an autoencoder ensemble via minority voting to differentiate between normal—always on—variables and production variables that may yield a higher entropy. Our approach has been tested in a production-equal smart factory and was cross-validated by a domain expert.

show abstract

“…Lei Le (Le, 2018) came up with the notion of a supervised autoencoders and showed empirically as well as theoretically that the architecture stacked across multiple hidden layers with different activation functions can improve generalizability of any network. In another interesting research by Abdulrahman Al-Abassi et al (Al-Abassi et al, 2020) the authors have tackled the data balancing issue in an imbalance data in real power systems while trying to detect the cyber-attacks. A deep learning ensemble method is proposed by fetching samples from imbalance datasets and sending it to a stacked auto encoder and deriving multiple balanced representations of the data.…”

Section: Literature Reviewmentioning

confidence: 99%

Enhanced Feature Extraction Using Optimal Whitening and Stacked Autoencoders

2022

Proceedings of the International Conferences on Applied Computing 2022 and WWW/Internet 2022

View full text Add to dashboard Cite

There are multiple preprocessing techniques used across the globe to enhance classification performance. Researchers in the past have incorporated various methods to improve feature extractions from image and text dataset. Extracting optimized features is still a challenge in the current scenarios, not only because size of the data is humongous, but it also involves noisy components. In this manuscript, we have proposed a method to compute the most optimized features from datasets by first transforming the features into the optimal ZCA features which are later fed into a 2-layered Stacked Autoencoder (SA). The novelty of the manuscript lies in the idea of deriving optimal features by performing optimal whitening (instead of plain ZCA) which is followed by feature extraction via stacked autoencoders. The in-depth features learned from SA are then fed into a feed-forward neural network as a final step towards classification. We have experimented rigorously on 10 non-computer vision, 2 hyperspectral remote sensing data set, and 1 computer vision datasets and found that classification accuracy increases to 2-4% on an average across many datasets. Additionally, we have validated the feature extraction method across plain vanilla neural network architecture and found the performance to increase by at most 5% on a few datasets.

show abstract

Unsupervised Stacked Autoencoders for Anomaly Detection on Smart Cyber-physical Grids

Cited by 26 publications

References 21 publications

A Review of Machine Learning Approaches in Synchrophasor Technology

A Review of Machine Learning Approaches in Synchrophasor Technology

Autoencoder-Ensemble-Based Unsupervised Selection of Production-Relevant Variables for Context-Aware Fault Diagnosis

Enhanced Feature Extraction Using Optimal Whitening and Stacked Autoencoders

Contact Info

Product

Resources

About