Browsing by Author "Leung, Ho Hon"
Now showing 1 - 11 of 11
Results Per Page
Sort Options
Item Autoencoder-based Intrusion Detection System(Institute of Electrical and Electronics Engineers Inc., 2021) Kamalov, Firuz; Zgheib, Rita; Leung, Ho Hon; Al-Gindy, Ahmed; Moussa, SherifGiven the dependence of the modern society on networks, the importance of effective intrusion detection systems (IDS) cannot be underestimated. In this paper, we consider an autoencoder-based IDS for detecting distributed denial of service attacks (DDoS). The advantage of autoencoders over traditional machine learning methods is the ability to train on unlabeled data. As a result, autoencoders are well-suited for detecting unknown attacks. The key idea of the proposed approach is that anomalous traffic flows will have higher reconstruction loss which can be used to flag the intrusions. The results of numerical experiments show that the proposed method outperforms benchmark unsupervised algorithms in detecting DDoS attacks. © 2021 IEEE.Item Deep learning regularization in imbalanced data(Institute of Electrical and Electronics Engineers Inc., 2020-11-03) Kamalov, Firuz; Leung, Ho HonDeep neural networks are known to have a large number of parameters which can lead to overfitting. As a result various regularization methods designed to mitigate the model overfitting have become an indispensable part of many neural network architectures. However, it remains unclear which regularization methods are the most effective. In this paper, we examine the impact of regularization on neural network performance in the context of imbalanced data. We consider three main regularization approaches: L{1}, L{2}, and dropout regularization. Numerical experiments reveal that the L{1} regularization method can be an effective tool to prevent overfitting in neural network models for imbalanced data. Index Terms-regularization, neural networks, imbalanced data. © 2020 IEEE.Item Divided difference operators in equivariant KK-theory(World Scientific Publishing Co. Pte. Ltd., 2014) Leung, Ho HonLet G be a compact connected Lie group with a maximal torus T. Let A, B be G-C*-algebras. We define certain divided difference operators on Kasparov's T-equivariant KK-group KKT(A, B) and show that KK G(A, B) is a direct summand of KKT(A, B). More precisely, a T-equivariant KK-class is G-equivariant if and only if it is annihilated by an ideal of divided difference operators. This result is a generalization of work done by Atiyah, Harada, Landweber and Sjamaar. © 2014 World Scientific Publishing Company.Item Ensemble Learning with Resampling for Imbalanced Data(Springer Science and Business Media Deutschland GmbH, 2021) Kamalov, Firuz; Elnagar, Ashraf; Leung, Ho HonImbalanced class distribution is an issue that appears in various applications. In this paper, we undertake a comprehensive study of the effects of sampling on the performance of bootstrap aggregating in the context of imbalanced data. Concretely, we carry out a comparison of sampling methods applied to single and ensemble classifiers. The experiments are conducted on simulated and real-life data using a range of sampling methods. The contributions of the paper are twofold: i) demonstrate the effectiveness of ensemble techniques based on resampled data over a single base classifier and ii) compare the effectiveness of different resampling techniques when used during the bagging stage for ensemble classifiers. The results reveal that ensemble methods overwhelmingly outperform single classifiers based on resampled data. In addition, we discover that NearMiss and random oversampling (ROS) are the optimal sampling algorithms for ensemble learning. © 2021, Springer Nature Switzerland AG.Item Equivariant structure constants for hamiltonian-T -spaces(TUBITAK, 2014) Leung, Ho HonIf there exists a set of canonical classes on a compact Hamiltonian-T -space in the sense of R Goldin and S Tolman, we derive some formulas for certain equivariant structure constants in terms of other equivariant structure constants and the values of canonical classes restricted to some fixed points. These formulas can be regarded as a generalization of Tymoczko's results. © TÜBİTAK.Item Feature Selection in Imbalanced Data(Springer Science and Business Media Deutschland GmbH, 2023-12) Kamalov, Firuz; Thabtah, Fadi; Leung, Ho HonThe traditional feature selection methods are not suitable for imbalanced data as they tend to be biased towards the majority class. This problem is particularly acute in the field of medical diagnostics and fraud detection where the class distribution is highly skewed. In this paper, we propose a novel filter approach using decision tree-based F1-score. The F1-score incorporates the accuracy with respect to the minority class data and hence is a good measure in the case of imbalanced data. In the proposed implementation, the F1-score is calculated based on a 1-dimensional decision tree classifier resulting in a fast and effective feature evaluation method. Numerical experiments confirm that the proposed method achieves robust dimensionality reduction and accuracy results. In addition, the low computational complexity of the algorithm makes it a practical choice for big data applications. © 2021, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.Item Leveraging computer algebra systems in calculus: A case study with SymPy(IEEE Computer Society, 2023) Kamalov, Firuz; Santandreu, David; Leung, Ho Hon; Johnson, Jason; El Khatib, ZiadItem Monotonicity of the χ2-statistic and Feature Selection(Springer Science and Business Media Deutschland GmbH, 2022) Kamalov, Firuz; Leung, Ho Hon; Moussa, SherifItem A note on property T for C * -algebras(Hikari Ltd., 2014) Kamalov, Firuz; Leung, Ho HonIn this note we introduce an alternative denition of Property T for C * -algebras based on the spectrum of a C * -algebra. We show that a group G has Property T if and only if C * r (G) has Property T. In addition, we introduce and investigate relative Property T for C * -algebras. © 2014 Firuz Kamalov and Ho Hon Leung.Item Outlier Detection in High Dimensional Data(World Scientific Publishing Co. Pte Ltd, 2020-03-01) Kamalov, Firuz; Leung, Ho HonHigh-dimensional data poses unique challenges in outlier detection process. Most of the existing algorithms fail to properly address the issues stemming from a large number of features. In particular, outlier detection algorithms perform poorly on dataset of small size with a large number of features. In this paper, we propose a novel outlier detection algorithm based on principal component analysis and kernel density estimation. The proposed method is designed to address the challenges of dealing with high-dimensional data by projecting the original data onto a smaller space and using the innate structure of the data to calculate anomaly scores for each data point. Numerical experiments on synthetic and real-life data show that our method performs well on high-dimensional data. In particular, the proposed method outperforms the benchmark methods as measured by F1-score. Our method also produces better-than-average execution times compared with the benchmark methods. © 2020 World Scientific Publishing Co.Item ROC curve model under Pareto distribution(Hikari Ltd., 2016) Leung, Ho Hon; Kamalov, FiruzThe receiver operating characteristic (ROC) curve is a useful graph-ical tool for analyzing the performance of a binary classifier. The area under the ROC curve (AUC) is a scalar measure of the classifier's per-formance. In this note we analyze the ROC curve derived under the assumption that the class distributions follow the Pareto model. We derive the equation of the ROC curve and compute the corresponding AUC. In addition, we discuss the optimal threshold for the classifier performance. © 2016 Firuz Kamalov and Ho Hon Leung.