Sequential Attack Detection in Recommender Systems

Aktukmak, Mehmet; Yılmaz, Yasin; Uysal, Ismail

doi:10.1109/tifs.2021.3076295

Cited by 6 publications

(5 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Specifically, unsupervised learning algorithms are widely used as they are highly accurate and fast by exploiting the nature of shilling profiles, and can be easily plugged into the CF framework [17]. For example, clustering-based techniques classify the abnormal ratings and normal ones [4], [5], and probability or inference-based techniques detect abnormal ratings with probability distribution [6], [7]. The work in [18] compares the clustering-based methods and probability-based methods.…”

Section: A Anomaly Detection In Recommender Systemsmentioning

confidence: 99%

“…To maintain the fairness and sustainability of recommender systems, previous research primarily concentrated on discovering the unusual rating patterns caused by shilling attacks. Most anomaly detection studies in CF-based recommender systems are for detecting malicious noise in the rating data, and all rating data without shilling attacks are considered normal data [4]- [7]. However, it should be noted that anomalies can exist unintentionally from a real user due to the user's psychological factors, which is called natural noise [2].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Autoencoder-Based Recommender System Exploiting Natural Noise Removal

Park

Jeong

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Collaborative filtering (CF) is a widely used technique in recommender systems by automatically predicting the user's latent interests based on many users' historical rating data. To improve the performance of the CF-based recommender systems, users' rating data should be pre-processed to avoid noise and enhance data reliability. Many researchers studied anomaly detection to remove malicious noise caused by shilling attacks, but anomalies can still exist in non-attacked real user data, which is called natural noise, as the ratings of users can be impacted by unpredictable factors such as other users' ratings and anchoring bias. In this paper, we propose an autoencoder-based recommendation system for exploiting the ability of both anomaly detection and CF. The proposed system detects the natural noise in the rating data based on the reconstruction errors after training. By removing the detected natural noise, CF can predict the unrated ratings with noise-free data. Our experiments show that the proposed model showed better performance than the traditional method by reducing the error by up to 5% compared to the method that does not consider natural noise detection and reducing the error by up to 4% compared to the conventional rating classification based natural noise detection methods.

show abstract

Section: A Anomaly Detection In Recommender Systemsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Autoencoder-Based Recommender System Exploiting Natural Noise Removal

Park

Jeong

et al. 2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The last term of the estimation error covariance P(k), as in (30), is the only term of ( 29) that depends on the attack; thus, maximizing the trace of the estimation error covariance is equivalent to maximizing the trace of its last term [50]. The last term of P(k) in ( 29) also depends on the selection matrix S(k) and given the attack covariance Σ, we can show that…”

Section: Coordinated Byzantine Attack Designmentioning

confidence: 99%

“…It is also vital to analyze how countermeasures taken by agents can reduce the impact of the attack. One approach is to detect adversaries and then implement corrective measures [28]- [30]. In [31], for example, attack detection is achieved through trusted agents that raise a flag when an adversary is detected.…”

Section: Introductionmentioning

confidence: 99%

Total Variation-Based Distributed Kalman Filtering for Resiliency Against Byzantines

2023

View full text Add to dashboard Cite

This paper proposes a Byzantine-resilient consensusbased distributed filter (BR-CDF) wherein network agents employ partial sharing of state parameters. We characterize the performance and convergence of the BR-CDF and study the impact of a coordinated data falsification attack. Our analysis shows that sharing merely a fraction of the states improves robustness against coordinated Byzantine attacks. In addition, we model the optimal attack strategy as an optimization problem where Byzantine agents design their attack covariance or the sequence of shared fractions to maximize the network-wide mean squared error (MSE). Numerical results demonstrate the accuracy of the proposed BR-CDF and its robustness against Byzantine attacks. Furthermore, the simulation results show that the influence of the covariance design is more pronounced when agents exchange larger portions of their states with neighbors. In contrast, the performance is more sensitive to the sequence of shared fractions when smaller portions are exchanged.

show abstract

“…Malicious software may be transmitted if the attacker is aware of an application vulnerability, such as a SQL injection or false data injection, according to [20] coding that results. [21] created a three-layer IDS for smart homes using a classification algorithm.…”

Section: Literature Reviewmentioning

confidence: 99%

A Novel Ensemble Model Using Learning Classifiers to Enhance Malware Detection for Cyber Security Systems

Das

Pramod

2022

IJRITCC

View full text Add to dashboard Cite

In the Internet of Things arena, smart gadgets are employed to offer quick and dependable access to services. IoT technology has the ability to recognize extensive information, provide information reliably, and process that information intelligently. Data networks, controllers, and sensors are increasingly used in industrial systems nowadays. Attacks have increased as a result of the growth in connected systems and the technologies they employ. These attacks may interrupt international business and result in significant financial losses. Utilizing a variety of methods, including deep learning (DL) and machine learning (ML), cyber assaults have been discovered. In this research, we provide an ensemble staking approach to efficiently and quickly detect cyber-attacks in the IoT. The NSL, credit card, and UNSW information bases were the three separate datasets used for the experiments. The suggested novel combinations of ensemble classifiers are done better than the other individual classifiers from the base model. Additionally, based on the test outcomes, it could be concluded that all tree and bagging-based combinations performed admirably and that, especially when their corresponding hyperparameters are set properly, differences in performance across methods are not significant statistically. Additionally, compared to other comparable PE (Portable Executable) malware detectors that were published recently, the suggested tree-based ensemble approaches outperformed them.

show abstract

Sequential Attack Detection in Recommender Systems

Cited by 6 publications

References 36 publications

Autoencoder-Based Recommender System Exploiting Natural Noise Removal

Autoencoder-Based Recommender System Exploiting Natural Noise Removal

Total Variation-Based Distributed Kalman Filtering for Resiliency Against Byzantines

A Novel Ensemble Model Using Learning Classifiers to Enhance Malware Detection for Cyber Security Systems

Contact Info

Product

Resources

About