Fully Homomorphic encryption (FHE) has been gaining in popularity as an emerging means of enabling an unlimited number of operations in an encrypted message without decryption. A major drawback of ...FHE is its high computational cost. Specifically, a bootstrapping step that refreshes the noise accumulated through consequent FHE operations on the ciphertext can even take minutes of time. This significantly limits the practical use of FHE in numerous real applications.By exploiting the massive parallelism available in FHE, we demonstrate the first instance of the implementation of a GPU for bootstrapping CKKS, one of the most promising FHE schemes supporting the arithmetic of approximate numbers. Through analyzing CKKS operations, we discover that the major performance bottleneck is their high main-memory bandwidth requirement, which is exacerbated by leveraging existing optimizations targeted to reduce the required computation. These observations motivate us to utilize memory-centric optimizations such as kernel fusion and reordering primary functions extensively.Our GPU implementation shows a 7.02× speedup for a single CKKS multiplication compared to the state-of-the-art GPU implementation and an amortized bootstrapping time of 0.423us per bit, which corresponds to a speedup of 257× over a single-threaded CPU implementation. By applying this to logistic regression model training, we achieved a 40.0× speedup compared to the previous 8-thread CPU implementation with the same data.
Phishing website attacks are a type of cyber-attack in which perpetrators create fraudulent websites that mimic legitimate platforms, such as online banking or social media, with the intent of ...tricking unsuspecting users into divulging sensitive information. This includes passwords, credit card details, usernames, and other personal data. These phishing websites are designed to look authentic and often employ various techniques, such as URL spoofing, social engineering, and email or text message phishing, to lure victims into revealing their confidential information. Web apps are growing increasingly complex and difficult to identify at first glance, especially when they use encryption and obfuscation techniques. In order to effectively detect and stop phishing web applications from being uploaded to the server in real-time, machine learning must be developed. In addition to including analyses for the machine learning algorithms for identifying web application-based assaults, the study calibrates fresh analyses by executing machine learning algorithms and confirming the findings. The study uses unique and categorized results from a machine learning dataset. As per the outcomes obtained from experimental and comparative analyses of the applied classification algorithms, the random forest model demonstrated the highest accuracy, achieving an impressive rate of 96.89%, followed by the decision tree model at 94.57%, and Extreme Gradient Boosting (XG).
In arid and semi-arid areas, groundwater resource is one of the most important water sources by the humankind. Knowledge of groundwater distribution over space, associated flow and basic exploitation ...measures can play a significant role in planning sustainable development, especially in arid and semi-arid areas. Groundwater potential mapping (GWPM) fits in this context as the tool used to predict the spatial distribution of groundwater. In this research we tested four GIS-based models for GWPM, consisting of: i) random forest (RF); ii) weight of evidence (WoE); iii) binary logistic regression (BLR); and iv) technique for order preference by similarity to ideal solution (TOPSIS) multi-criteria. The Shahroud plain located in Iran, was selected to research the water scarcity and over-exploitation of groundwater resources over the past 20 years. In this research, using Iranian Department of Water Resources Management data, and extensive field surveys, 122 groundwater well data with high potential yield of ≥11 m3 h−1 were selected for GWPM. Specifically, we generated four different models selecting 70% (n = 85) of the wells and validated the resulting GWP maps upon the complementary 30% (n = 37).A total of fifteen ground water conditioning factors to explain the groundwater well distribution over the Shahroud plain were selected. From the Advanced Land Observing Satellite (ALOS), a DEM (30 m resolution) was extracted to calculate a set of morphometric properties which were combined with thematic ones such as land use/land cover (LU/LC) and Soil Type (ST). Results show that in RF (LU/LC), LR (ST), and AHP (Slope) are the most relevant contributors to groundwater occurrence. After that, using the natural break method, final maps were divided into five susceptibility classes of very low, low, moderate, high, and very high. The accuracy of models was ultimately tested using prediction rate (validation data), success rate (training data) and the seed cell area index (SCAI) indicators. Results of validation show that BLR with prediction rate of 0.905 (90.5%) and success rate of 0.918 (91.8%) had higher accuracy than WoE, RF and TOPSIS models with respective prediction rates of 0.885, 0.873 and 0.870 (88.5%, 87.3%, and 87%) and success rate of 0.900, 0.889, and 0.881 (90%, 88.9%, and 88.1%). SCAI results show that all models have acceptable classification accuracy although BLR outperformed the other models in terms of accuracy. Results show that the combination of remote sensing (RS) data and geographic information system (GIS) with new approaches can be used as a powerful tool in GWPM in arid and semi-arid areas. The results of this investigation introduced a potential novel methodology that could be used by decision-makers for the sustainable management of ground water resources.
Display omitted
•Four bivariate and multivariate statistical (WOE and BLR), data mining (RF) and multi criteria decision making (TOPSIS) models were applied for groundwater potential mapping.•The capability and robustness of all models were assessed using AUC and BLR with prediction rate of 0.905 and success rate of 0.918 had higher accuracy than WOE, RF and TOPSIS models.•Land use/land cover, soil type, and slope were the key factors in groundwater occurrence in RF, BLR and AHP respectively.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
Quantitative assessment and long-term analysis of ecological vulnerability can effectively grasp the driving factors of environmental change, which is of great significance for ecological protection ...and restoration. This study took 381 villages in Jingle County, a typical ecological vulnerable area on the Loess Plateau, as basic evaluation unit, and an ecological vulnerability evaluation index system with 12 evaluation indexes was constructed based on “sensitivity-resilience-pressure” (SRP) conceptual model. Combined with spatial principal component analysis, spatial autocorrelation, cross-sectional and panel regression, the spatio-temporal variation characteristics and driving factors of ecological vulnerability in 2005, 2010 and 2015 were quantitatively analyzed. The results showed that areas classified as not and somewhat vulnerable were expanding, while areas classified as highly and extremely vulnerable were shrinking from 2005 to 2015. The variations of ecological vulnerability were the result of interaction of natural environment and human activities, among which Normalized Difference Vegetation Index (NDVI), distance from main road and distance from river system were conducive to reducing ecological vulnerability, moreover, NDVI has the greatest impact on the probability of ecological vulnerability transition from high to low, with a regression coefficient of 46.66. The impact of social factors decreased relatively, while the role of natural factors increased, and increasing vegetation coverage and economic development helped reduce ecological vulnerability.
Display omitted
•The impact of social factors on ecological vulnerability was declining relatively.•Natural factors had an increasing role on ecological vulnerability with time.•Ecological protection and restoration reduced the area's ecological vulnerability.•Ecological vulnerability evaluation can guide ecological restoration practice.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
The aim of the study is to select adequate econometric instruments for building a scoring model on a specific array of initial data, which contains the vast majority of fictitious variables. Despite ...a significant number of developments devoted to the construction of scoring models, a universal method allowing to obtain a highly efficient classifier for any data has not been identified. Therefore, the task of selection of the best method for building a scoring model remains relevant, depending on the characteristics of the available data. The most successful approach when selecting a model for solving the problem of binary classification is the use of several types of econometric models and the choice of the best of them according to the results of classification. In the presented study, the following types of models were applied: discriminant model, logit and probit regressions, and polynomial logistic regression. Training samples with different structure were used. Comparison of all obtained models allows us to conclude that polynomial logistic regression is preferable in this case. This model demonstrates high classification rates for all introduced object classes and has an important advantage compared to models that make a binary selection. The advantage of polynomial logistic regression is also the possibility of selecting in each case a convenient scale for dividing borrowers into more than two classes and determining the level of probability of reliability of the borrower acceptable for its own conditions, at which it should be assigned to one of the selected classes. Prospects for further research in this direction are the use of machine learning methods that will be able to use ensembles of the best of the considered models. In addition, the proposed models can be used in solving similar problems in other spheres of economic activity.
•Public transport vehicles could be hotspots for virus transmission during pandemics.•A binary logistic regression model was developed to explain mode choice during COVID-19.•Compared to males ...females are more likely to use public transport relative to solo modes.•People with higher trip frequency are more likely to use public transport relative to solo modes.•People with higher income levels tend to use public transport less often.•Age was not a significant predictor of mode choices.
A sharp decline in public transport use has been reported worldwide since the outbreak of the COVID-19 pandemic. As the virus spreads through close contacts, particularly in closed environments, public transport vehicles could be considered as hotspots for its transmission. However, public transport operations cannot be entirely stopped as many people in developing countries rely on them for their travel needs. This study aims to provide insights into people's travel mode choices during the COVID-19 pandemic. Data, i.e., 1,516 complete survey responses, were obtained through a questionnaire that was conducted in Lahore, Pakistan. A binary logistic model was developed using the collected data to model the likelihood of choosing solo or public transport modes during COVID-19. The results explained that the respondents preferred solo modes more than the public transport modes during the pandemic. Gender, income, education, profession, trip frequency, car ownership, motorbike ownership, and an underlying factor that was defined as “safety precautions” were found to be significant predictors of the public transport choice relative to solo modes. Females tend to choose public transport modes relative to solo modes as compared to males. Private vehicle (car or motorbike) owners were less likely to use public modes relative to solo modes when compared to those who do not own private vehicles. The outcomes of this study could be important for the government authorities, policymakers, and transport operators to understand the public transport use in developing countries during pandemics. Such information will be useful to devise regulations and preventive measures to control infectious diseases associated with public transport use, particularly in developing countries, where private transport options are limited.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
The present study explores one of the biggest causes of the lack of organizational knowledge creation: knowledge hiding (KH). KH can be provoked by the deviant and detached behaviours of leaders ...and/or the motivations of employees. In this context, leaders assume a key role in reducing the effect of KH. Through the lens of transformational leadership (TL), a sample of 758 European SMEs with a total number of 2,232 employees operating in a knowledge-intensive sector is investigated. The scope is to evaluate the correlation between the three main characteristics of transformational leadership (i.e., trust, a collaborative environment, and the involvement of employees) and the phenomenon of KH through a logistic regression analysis. It emerges that TL can influence the organizational context and redefine the behaviours related to KH. In addition, empathic leadership can provide added value for companies since a collaborative environment and common objectives reduce the level of KH.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NLZOH, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UILJ, UL, UM, UPCLJ, UPUK, ZAGLJ, ZRSKP
This paper investigates the feature subset selection problem for the binary classification problem using logistic regression model. We developed a modified discrete particle swarm optimization (PSO) ...algorithm for the feature subset selection problem. This approach embodies an adaptive feature selection procedure which dynamically accounts for the relevance and dependence of the features included the feature subset. We compare the proposed methodology with the tabu search and scatter search algorithms using publicly available datasets. The results show that the proposed discrete PSO algorithm is competitive in terms of both classification accuracy and computational performance.
Full text
Available for:
GEOZS, IJS, IMTLJ, KILJ, KISLJ, NUK, OILJ, PNG, SAZU, SBCE, SBJE, UL, UM, UPCLJ, UPUK
Background
Preventive measures to decrease the frequency and intensity of anaphylactic events are essential to provide optimal care for allergic patients. Aggravating factors may trigger or increase ...the severity of anaphylaxis and therefore need to be recognized and avoided.
Objective
To identify and prioritize factors associated with an increased risk of developing severe anaphylaxis.
Methods
Data from the Anaphylaxis Registry (122 centers in 11 European countries) were used in logistic regression models considering existing severity grading systems, elicitors, and symptoms to identify the relative risk of factors on the severity of anaphylaxis.
Results
We identified higher age and concomitant mastocytosis (OR: 3.1, CI: 2.6‐3.7) as the most important predictors for an increased risk of severe anaphylaxis. Vigorous physical exercise (OR: 1.5, CI: 1.3‐1.7), male sex (OR: 1.2, CI: 1.1‐1.3), and psychological burden (OR: 1.4, CI: 1.2‐1.6) were more often associated with severe reactions. Additionally, intake of beta‐blockers (OR: 1.9, CI: 1.5‐2.2) and ACE‐I (OR: 1.28, CI: 1.05, 1.51) in temporal proximity to allergen exposition was identified as an important factor in logistic regression analysis.
Conclusion
Our data suggest it may be possible to identify patients who require intensified preventive measures due to their relatively higher risk for severe anaphylaxis by considering endogenous and exogenous factors.
Full text
Available for:
BFBNIB, FZAB, GIS, IJS, KILJ, NLZOH, NUK, OILJ, SAZU, SBCE, SBMB, UL, UM, UPUK