Phishing Website Detection Based on Deep Convolutional Neural Network and Random Forest Ensemble Learning

Phishing has become one of the biggest and most effective cyber threats, causing hundreds of millions of dollars in losses and millions of data breaches every year. Currently, anti-phishing techniques require experts to extract phishing sites features and use third-party services to detect phishing sites. These techniques have some limitations, one of which is that extracting phishing features requires expertise and is time-consuming. Second, the use of third-party services delays the detection of phishing sites. Hence, this paper proposes an integrated phishing website detection method based on convolutional neural networks (CNN) and random forest (RF). The method can predict the legitimacy of URLs without accessing the web content or using third-party services. The proposed technique uses character embedding techniques to convert URLs into fixed-size matrices, extract features at different levels using CNN models, classify multi-level features using multiple RF classifiers, and, finally, output prediction results using a winner-take-all approach. On our dataset, a 99.35% accuracy rate was achieved using the proposed model. An accuracy rate of 99.26% was achieved on the benchmark data, much higher than that of the existing extreme model.

Exploring the Knowledge Embedded in Class Visualizations and Their Application in Dataset and Extreme Model Compression

Artificial neural networks are efficient learning algorithms that are considered to be universal approximators for solving numerous real-world problems in areas such as computer vision, language processing, or reinforcement learning. To approximate any given function, neural networks train a large number of parameters—up to millions, or even billions in some cases. The large number of parameters and hidden layers in neural networks make them hard to interpret, which is why they are often referred to as black boxes. In the quest to make artificial neural networks interpretable in the field of computer vision, feature visualization stands out as one of the most developed and promising research directions. While feature visualizations are a valuable tool to gain insights about the underlying function learned by the network, they are still considered to be simple visual aids requiring human interpretation. In this paper, we propose that feature visualizations—class visualizations in particular—are analogous to mental imagery in humans, resembling the experience of seeing or perceiving the actual training data. Therefore, we propose that class visualizations contain embedded knowledge that can be exploited in a more automated manner. We present a series of experiments that shed light on the nature of class visualizations and demonstrate that class visualizations can be considered a conceptual compression of the data used to train the underlying model. Finally, we show that class visualizations can be regarded as convolutional filters and experimentally show their potential for extreme model compression purposes.

A Flexible Extension to an Extreme Distribution

The aim of this paper is not only to propose a new extreme distribution, but also to show that the new extreme model can be used as an alternative to well-known distributions in the literature to model various kinds of datasets in different fields. Several of its statistical properties are explored. It is found that the new extreme model can be utilized for modeling both asymmetric and symmetric datasets, which suffer from over- and under-dispersed phenomena. Moreover, the hazard rate function can be constant, increasing, increasing–constant, or unimodal shaped. The maximum likelihood method is used to estimate the model parameters based on complete and censored samples. Finally, a significant amount of simulations was conducted along with real data applications to illustrate the use of the new extreme distribution.

Historical Civil Registration Record Transcription Using an eXtreme Model Driven Approach

Modelling is considered as a universal approach to define and simplify real-world applications through appropriate abstraction. Model-driven system engineering identifies and integrates appropriate concepts, techniques, and tools which provide important artefacts for interdisciplinary activities. In this paper, we show how we used a model-driven approach to design and improve a Digital Humanities dynamic web application within an interdisciplinary project that enables history students and volunteers of history associations to transcribe a large corpus of image-based data from the General Register Office (GRO) records. Our model-driven approach generates the software application from data, workflow and GUI abstract models, ready for deployment.

Historical Civil Registration Record Transcription Using an eXtreme Model Driven Approach

Modelling is considered as a universal approach to define and simplify real-world applications through appropriate abstraction. Model-driven system engineering identifies and integrates appropriate concepts, techniques, and tools which provide important artefacts for interdisciplinary activities. In this paper, we show how we used a model-driven approach to design and improve a Digital Humanities dynamic web application within an interdisciplinary project that enables history students and volunteers of history associations to transcribe a large corpus of image-based data from the General Register Office (GRO) records. Our model-driven approach generates the software application from data, workflow and GUI abstract models, ready for deployment.

Pre-Competition Weight Loss Models in Taekwondo: Identification, Characteristics and Risk of Dehydration

Athletes use different combinations of weight loss methods during competition preparation. The aim of this study was to identify and characterize pre-competition weight loss models, which describe these combinations. The second aim was to determine if any existing model pose a higher risk of severe dehydration and whether any of the models could be continued as a lower-risk option. The third aim was to explore whether athletes who used different weight management strategies could be differentiated based on age, sex, training experience or anthropometric parameters. Study participants were randomly selected from Olympic taekwondo competitors and 192 athletes were enrolled. Active (47% weight-reducing athletes), passive (31%) and extreme (22%) models have been described. In the extreme model, athletes combined the highest number of different weight loss methods (3.9 ± 0.9 methods vs. 2.4 ± 0.9 in active and 1.5 ± 0.6 in passive), reduced significantly more body mass than others (6.7 ± 3.5% body mass vs. 4.3 ± 1.9% and 4.5 ± 2.4%; p < 0.01) and all of them used methods with the highest risk of severe dehydration. The active and passive models could be continued as a lower-risk option, if athletes do not combine dehydrating methods and do not prolong the low energy availability phase. The extreme model carried the highest risk of severe dehydration. Every fifth weight-reducing taekwondo athlete may have been exposed to the adverse effects of acute weight loss. Taekwondo athletes, regardless of age, sex, training experience and anthropometric parameters, lose weight before the competition and those characteristics do not differentiate them between models.

Exploring the dispersion measure of the Milky Way halo

Fast radio bursts offer the opportunity to place new constraints on the mass and density profile of hot and ionized gas in galactic haloes. We test here the X-ray emission and dispersion measure predicted by different gas profiles for the halo of the Milky Way. We examine a range of models, including entropy stability conditions and external pressure continuity. We find that incorporating constraints from X-ray observations leads to favouring dispersion measures on the lower end of the range given by these models. We show that the dispersion measure of the Milky Way halo could be less than 10 cm−3 pc in the most extreme model we consider, which is based on constraints from O vii absorption lines. However, the models allowed by the soft X-ray constraints span more than an order of magnitude in dispersion measures. Additional information on the distribution of gas in the Milky Way halo could be obtained from the signature of a dipole in the dispersion measure of fast radio bursts across the sky, but this will be a small effect for most cases.