1,158 Hits in 7.5 sec

Self-Adaption AAE-GAN for Aluminum Electrolytic Cell Anomaly Detection

Danyang Cao, Di Liu, Xu Ren, Nan Ma
2021 IEEE Access  
In this article, we use the ability of GAN to model complex high-dimensional image distribution, and propose a self-adaption AAE-GAN network based on adaptive changes of input samples.  ...  In recent years, generative adversarial network (GAN) has become more and more popular in the field of anomaly detection.  ...  based on the algorithm direction [49] - [51] proposed two methods based on data direction, they used sampling strategies, such as under-sampling and over-sampling techniques to improve the problem  ... 
doi:10.1109/access.2021.3097116 fatcat:ufxi7wohc5hjjd7esyv3uksd5q

On the Performance of Generative Adversarial Network by Limiting Mode Collapse for Malware Detection Systems

Acklyn Murray, Danda B. Rawat
2021 Sensors  
In this paper, we study conditional limiter solutions for mode collapse for the Intrusion Detection System (IDS) Control Flow GAN (ICF-GAN) model.  ...  the input data.  ...  Our goal is to study to minimize mode collapse in a GAN framework purposed for malware detection based on PCAP data where data ingestion to the discriminator in GAN impacts the data modifications, control  ... 
doi:10.3390/s22010264 pmid:35009810 pmcid:PMC8749644 fatcat:kl3uiq7ykbbcvcc6iciunfiqxa

STEP-GAN: A Step-by-Step Training for Multi Generator GANs with application to Cyber Security in Power Systems [article]

Mohammad Adiban, Arash Safari, Giampiero Salvi
2020 arXiv   pre-print
In this study, we introduce a novel unsupervised countermeasure for smart grid power systems, based on generative adversarial networks (GANs).  ...  Our model simulates possible attacks on power systems using multiple generators in a step-by-step interaction with a discriminator in the training phase.  ...  S14 S15 Dataset The experiments are conducted on an open-source simulated ICS cyber attack dataset obtained from Supervisory Control and Data Acquisition (SCADA) power systems provided by Mississippi  ... 
arXiv:2009.05184v1 fatcat:exh5b336qffffdwa354hj3jvju

IMIDS: An Intelligent Intrusion Detection System against Cyber Threats in IoT

Kim-Hung Le, Minh-Huy Nguyen, Trong-Dat Tran, Ngoc-Duan Tran
2022 Electronics  
Furthermore, IMIDS's detection performance is notably improved after being further trained by the data generated by our attack data generator.  ...  To mitigate the training data shortage issue, we also propose an attack data generator powered by a conditional generative adversarial network.  ...  Indeed, traditional GAN generators are fed by random data samples from the dataset, so they are inefficient with imbalanced datasets.  ... 
doi:10.3390/electronics11040524 doaj:d0f7becc6b6a4f82ae419f669842eed5 fatcat:2ldpmofijrbqzfoj5ckotqnffu

HML-IDS: A Hybrid-Multilevel Anomaly Prediction Approach for Intrusion Detection in SCADA Systems

Izhar Ahmed Khan, Dechang Pi, Zaheer Ullah Khan, Yasir Hussain, Asif Nawaz
2019 IEEE Access  
., electricity generation and dispersal networks, chemical processing plants, and gas distribution, are governed and monitored by supervisory control and data acquisition systems (SCADA).  ...  The challenge in constructing an intrusion detection framework is to deal with unbalanced intrusion datasets, i.e. when one class is signified by a lesser amount of instances (minority class).  ...  They build their model by training only from normal flow of traffic data. In an attempt to detect novelty, the authors of [42] also proposed one-class classification method (ALOCC) based on GANs.  ... 
doi:10.1109/access.2019.2925838 fatcat:lnyhxoajv5hkhjso4pyae7vbii

Synthetic Data – what, why and how? [article]

James Jordon, Lukasz Szpruch, Florimond Houssiau, Mirko Bottarelli, Giovanni Cherubin, Carsten Maple, Samuel N. Cohen, Adrian Weller
2022 arXiv   pre-print
This explainer document aims to provide an overview of the current state of the rapidly expanding work on synthetic data technologies, with a particular focus on privacy.  ...  We do believe that synthetic data is a very useful tool, and our hope is that this report highlights that, while drawing attention to nuances that can easily be overlooked in its deployment.  ...  Acknowledgements We would like to thank Accenture, Hazy, HSBC, MOSTLY AI, and the Office for National Statistics for their participation in our interviews.  ... 
arXiv:2205.03257v1 fatcat:2sdjqyfdivb4hnxn3jr2j5e6q4

SingleNet: A Lightweight Convolutional Neural Network for Safety Detection of an Industrial Control System

Yun Sha, Jianping Chen, Jianwang Gan, Yong Yan, Xuejun Liu, Hao Wang, Hye-jin Kim
2022 Mobile Information Systems  
, model size, and iteration time on the industrial control datasets.  ...  In this paper, a lightweight convolutional neural network anomaly detection algorithm "SingleNet" suitable for the edge side of the industrial control system is proposed, which convolutes the data of each  ...  of data to detect the physical and network attacks on industrial control systems.  ... 
doi:10.1155/2022/1148518 fatcat:zuufagx6qnddhidjzmrhihtmkq

ARIES: A Novel Multivariate Intrusion Detection System for Smart Grid

Panagiotis Radoglou Grammatikis, Panagiotis Sarigiannidis, Georgios Efstathopoulos, Emmanouil Panaousis
2020 Sensors  
By emphasising on the third layer, the ARIES Generative Adversarial Network (ARIES GAN) with novel error minimisation functions was developed, considering mainly the reconstruction difference.  ...  Based on the evaluation analysis, the proposed GAN network overcomes the efficacy of conventional ML methods in terms of Accuracy and the F1 score.  ...  The anomalous records were identified by specific experts. A sample of this dataset is illustrated in Figure 5 .  ... 
doi:10.3390/s20185305 pmid:32948064 pmcid:PMC7570496 fatcat:bwl6ok7opza6vkusamkti7yfty

Towards digital cognitive clones for the decision-makers: adversarial training experiments

Mariia Golovianko, Svitlana Gryshko, Vagan Terziyan, Tuure Tuunanen
2021 Procedia Computer Science  
This enables virtual presence of a professional decision-maker simultaneously in many places and processes of Industry 4.0.  ...  This enables virtual presence of a professional decision-maker simultaneously in many places and processes of Industry 4.0.  ...  It would allow collective intelligence (integrated digital customers and humans) interacting with real services and products via their digital twins in cyber-physical environments.  ... 
doi:10.1016/j.procs.2021.01.155 fatcat:74qe4vvcurbftco53z45baupoq

Differential Privacy: What is all the noise about? [article]

Roxana Danger
2022 arXiv   pre-print
This paper aims to provide an overview of the most important ideas, concepts and uses of DP in ML, with special focus on its intersection with Federated Learning (FL).  ...  Differential Privacy (DP) is a formal definition of privacy that provides rigorous guarantees against risks of privacy breaches during data processing.  ...  : 5 Figure 5: General architecture of a GAN Algorithm 3 :••145 9 Generate another D G 10 Go Step 1 11 39101 PATE-GAN Input: Training data: D ∪ D G where D G is a dataset generated by using the public  ... 
arXiv:2205.09453v1 fatcat:5z3nqsh7qbbwfhbrc6hmzt43ya

Smart Grid Security and Privacy: From Conventional to Machine Learning Issues (Threats and Countermeasures)

Parya Haji Mirzaee, Mohammad Shojafar, Haitham Cruickshank, Rahim Tafazolli
2022 IEEE Access  
However, despite these algorithms' high accuracy and reliability, ML systems are also vulnerable to a group of malicious activities called adversarial ML (AML) attacks.  ...  Besides, with the real-time information flow, and online energy consumption controlling systems, customers' privacy and preserving their confidential data in SG is critical to be addressed.  ...  In this method, attackers are categorised based on their knowledge of the training dataset, learning algorithm, and samples [115] .  ... 
doi:10.1109/access.2022.3174259 fatcat:txuebjhpnre73cq5lbx77ugmhq

Internet of Things Applications, Security Challenges, Attacks, Intrusion Detection, and Future Visions: A Systematic Review

Nivedita Mishra, Sharnil Pandya
2021 IEEE Access  
Furthermore, the classification of Intrusion Detection Systems, different anomaly detection techniques, different Intrusion Detection System models based on datasets, various machine learning and deep  ...  Different types of DDoS attacks, DDoS attacks in IoT devices, impacts of DDoS attacks, and solutions for mitigation are discussed in detail.  ...  Attack-Network includes one router, one switch, and four PCs. This kind of network is generally used for data generation. Dataset generated in this work contains attacks based on McAfee report 2016.  ... 
doi:10.1109/access.2021.3073408 fatcat:ebzvtidh2relplv3kn3t6plygu

Artificial Intelligence and Machine Learning in 5G Network Security: Opportunities, advantages, and future research trends [article]

Noman Haider, Muhammad Zeeshan Baig, Muhammad Imran
2020 arXiv   pre-print
Therefore, AI and ML can play central role in protecting highly data-driven softwareized and virtualized network components.  ...  Also, an overview of key data collection points in 5G architecture for threat classification and anomaly detection are discussed.  ...  By using generator and discriminator DNNs in a single training mechanism, the networks compete with one another where the generator generates new data samples whereas the discriminator distinguish them  ... 
arXiv:2007.04490v1 fatcat:wlpeaoyxbjc5pgfctx2eodwrpa

Generative Adversarial Networks (GANs) in networking: A comprehensive survey & evaluation

Hojjat Navidan, Parisa Fard Moshiri, Mohammad Nabati, Reza Shahbazian, Seyed Ali Ghorashi, Vahid Shah-Mansouri, David Windridge
2021 Computer Networks  
Despite the recency of their conception, Generative Adversarial Networks (GANs) constitute an extensively researched machine learning sub-field for the creation of synthetic data through deep generative  ...  Given their relative ease of use, it is therefore natural that researchers in the field of networking (which has seen extensive application of deep learning methods) should take an interest in GAN-based  ...  We train each of the five GANs on each of the datasets separately and use noise drawn from the same latent space to generate samples.  ... 
doi:10.1016/j.comnet.2021.108149 fatcat:4ekgil24ijha3evmzruez63tdq

Differential Privacy Preservation in Deep Learning: Challenges, Opportunities and Solutions

Jingwen Zhao, Yunfang Chen, Wei Zhang
2019 IEEE Access  
In this paper, we introduce the privacy attacks facing the deep learning model and present them from three aspects: membership inference, training data extraction, and model extracting.  ...  Nowadays, deep learning has been increasingly applied in real-world scenarios involving the collection and analysis of sensitive data, which often causes privacy leakage.  ...  Dalenius [11] proposed the concept of private disclosure control, and the k-anonymity algorithm [12] lays a foundation for the anonymous privacy protection algorithm based on equivalence class grouping  ... 
doi:10.1109/access.2019.2909559 fatcat:zgbo63onnzcqpmzjvh5mf45gke
« Previous Showing results 1 — 15 out of 1,158 results