Problem Domain Research Articles

Multimodal Religiously Hateful Social Media Memes Classification Based on Textual and Image Data

Multimodal hateful social media meme detection is an important and challenging problem in the vision-language domain. Recent studies show high accuracy for such multimodal tasks due to datasets that provide better joint multimodal embedding to narrow the semantic gap. Religiously hateful meme detection is not extensively explored among published datasets. While there is a need for higher accuracy on religiously hateful memes, deep learning–based models often suffer from inductive bias. This issue is addressed in this work with the following contributions. First, a religiously hateful memes dataset is created and published publicly to advance hateful religious memes detection research. Over 2000 meme images are collected with their corresponding text. The proposed approach compares and fine-tunes VisualBERT pre-trained on the Conceptual Caption (CC) dataset for the downstream classification task. We also extend the dataset with the Facebook hateful memes dataset. We extract visual features using ResNeXT-152 Aggregated Residual Transformations–based Masked Regions with Convolutional Neural Networks (R-CNN) and Bidirectional Encoder Representations from Transformers (BERT) uncased for textual encoding for the early fusion model. We use the primary evaluation metric of an Area Under the Operator Characters Curve (AUROC) to measure model separability. Results show that the proposed approach has a higher AUROC score of 78%, proving the model’s higher separability performance and an accuracy of 70%. It shows comparatively superior performance considering dataset size and against ensemble-based machine learning approaches.

Comparative Analysis of Nature-Inspired Metaheuristic Techniques for Optimizing Phishing Website Detection

The increasing number, frequency, and sophistication of phishing website-based attacks necessitate the development of robust solutions for detecting phishing websites to enhance the overall security of cyberspace. Drawing inspiration from natural processes, nature-inspired metaheuristic techniques have been proven to be efficient in solving complex optimization problems in diverse domains. Following these successes, this research paper aims to investigate the effectiveness of metaheuristic techniques, particularly Genetic Algorithms (GAs), Differential Evolution (DE), and Particle Swarm Optimization (PSO), in optimizing the hyperparameters of machine learning (ML) algorithms for detecting phishing websites. Using multiple datasets, six ensemble classifiers were trained on each dataset and their hyperparameters were optimized using each metaheuristic technique. As a baseline for assessing performance improvement, the classifiers were also trained with the default hyperparameters. To validate the genuine impact of the techniques over the use of default hyperparameters, we conducted statistical tests on the accuracy scores of all the optimized classifiers. The results show that the GA is the most effective technique, by improving the accuracy scores of all the classifiers, followed by DE, which improved four of the six classifiers. PSO was the least effective, improving only one classifier. It was also found that GA-optimized Gradient Boosting, LGBM and XGBoost were the best classifiers across all the metrics in predicting phishing websites, achieving peak accuracy scores of 98.98%, 99.24%, and 99.47%, respectively.

Are We All in This Together? Examining Nonprofits’ Perceptions of Governmental Actors in the Management of the U.S. Refugee Crisis

Abstract U.S. federal policy has created, at best, a gap and, at worst, a hostile environment for nonprofits serving refugees. We rely on frameworks of nonprofit-government relationships (institutional voids, structural holes, instrumental/expressive support) to explore government-nonprofit interactions in the refugee domain, and their impact on 34 refugee-serving nonprofits in the U.S. Findings indicate limited expressive and instrumental support for nonprofits and suggest nonprofits must navigate complex, multilevel, environments. Contributions include the suggestion of “intentional” rather than institutional voids, and a new typology of forms (zero, unclaimed, symbolic, or comprehensive) of government support for nonprofits in a problem domain based on whether government’s instrumental and expressive support for nonprofits is high or low.

Complete Solutions in the Dilatation Theory of Elasticity with a Representation for Axisymmetry

In this paper, we present certain complete solutions in the dilatation theory of elasticity. This model can be derived as a special case of Eringen’s linear theory of microstretched elastic solids when microrotations are absent. It is also a version of the theory of materials with voids. The dilatation theory can be considered the simplest theoretical model of microstructured materials and is suitable for investigating various phenomena that occur in engineering, geomechanics, and biomechanics. We establish three complete solutions to the displacement equations of equilibrium that are the counterpart of the Green–Lamé (GL), Boussinesq–Papkovich–Neuber (BPN), and Cauchy–Kowalevski–Somigliana (CKS) solutions of classical elasticity. The links between these BPN and CKS solutions are established. Then, we present a representation of the BPN solution in the case of axisymmetry. The results presented here are useful for solving axisymmetric problems in semi-infinite and infinite domains.

Tempered fractional Sobolev spaces

The fractional Laplacian operator Δs is the infinitesimal generator of the isotropic (2s)-stable Lévy process on Rn, which is the scaling limit of the Lévy flight with isotropic power law measure |x|−n−2s. However, their second and higher moments are divergent, leading to the difficulty in the modeling of practical physical processes. Replacing |x|−n−2s by the measure of an isotropic tempered power law with the tempering exponent λ (i.e., e−λ|x||x|−n−2s), the tempered fractional Laplacian operator (Δ+λ)s was introduced in [17] as the infinitesimal generator of the tempered Lévy process. In this paper, guided by the fractional Sobolev spaces Ws,p corresponding to the fractional Laplacian operator Δs, we deal with the tempered fractional Sobolev spaces Ws,λ,p associated with the tempered fractional Laplacian (Δ+λ)s. First, the definition of the tempered fractional Sobolev spaces Ws,λ,p is given via the Gagliardo approach, and some of their basic properties were studied. Subsequently, we focus on the Hilbert case Hs,λ based on the Fourier transform. In particular, we deal with its relation with the tempered fractional Sobolev space Ws,λ,2 and analyze their role in the trace theory, overcoming the challenges posed by the tempering exponent λ. Then we investigate the asymptotic behavior of λ→0+,s→1− and s→0+ that appear in the definition of the tempered fractional Laplacian operator (Δ+λ)s. Moreover, we show continuous and compact embeddings investigating the problem of the extension domains and the generalized Hölder regularity results. As an application of the tempered fractional Sobolev spaces, we prove that the process defined by the tempered Lévy process is a solution of some PDE with tempered fractional Lapalcian operator.

Existence of optimal domains for the helicity maximisation problem among domains satisfying a uniform ball condition

In the present work we present a general framework which guarantees the existence of optimal domains for isoperimetric problems within the class of C1,1-regular domains satisfying a uniform ball condition as long as the desired objective function satisfies certain properties. We then verify that the helicity isoperimetric problem studied in [Cantarella et al., J. Math. Phys. 41, 5615 (2000)] satisfies the conditions of our framework and hence establish the existence of optimal domains within the given class of domains. We additionally use the same framework to prove the existence of optimal domains among uniform C1,1-domains for a first curl eigenvalue problem which has been studied recently for other classes of domains in [Enciso et al., Trans. Am. Math. Soc. 377, 4519–4540 (2024)].

Enhancement of cooling process of hot blocks mounted inside a horizontal channel using flexible baffles — Alternative arrangement

In this paper, two heated blocks and three flexible baffles that are alternately placed on the upper and bottom sides of a rectangular channel are used to replicate the transient heat and air movement. The novelty of this research is the use of flexible baffles to destruct the thermal boundary layer with maintaining little pressure drop, and thus reducing the required pumping capacity. The finite thickness blocks are subject to a constant heat flux, while the baffles are very thin and deforms depending on the strength of the flow of coolant (air). Finite element method with arbitrary Lagrangian–Eulerian approach is used for solving such deforming domain problem. The problem is inspected transiently by varying the dimensionless time (τ=0.0 to 10) and two salient parameters which are Reynolds number Re=10−500, and baffle elasticity parameter E=105−108. Results show that the flexibility feature of the baffles contributes not only in directing the coolant towards the hot blocks, but also contributes in reducing the skin friction and pressure drop through the channel. At Re=150, the skin friction and pressure drop of the flexible baffles are 13.2% and 16.5%, respectively lesser than the rigid baffles. The improvement of Nusselt number with flexibility parameter is marginal and within a limited low Reynolds number, where the maximal improvement is about 3.4% at Re=150.

Continuum and Discrete Analytical Methods for Vibration and Buckling Eigenvalues Shape Sensitivities

AbstractGradient-based optimization techniques need precise and efficient sensitivities. Numerical sensitivity methods such as finite differences are easy to implement but imprecise and computationally inefficient. In contrast, analytical sensitivity methods, such as the discrete and continuum ones are highly accurate and efficient. Continuum Sensitivity Analysis (CSA) is an analytical method used to calculate derivatives of shape or value parameters. While CSA has been successfully applied in static analysis and dynamic problems in the time domain, this work presents an extension of the approach to eigenvalue sensitivities for the first time. However, CSA revealed limitations, prompting the exploration of an alternative approach based on discrete analytical differentiation. This method is employed for the first time in shape sensitivities. The derivatives of the stiffness and mass matrices required by the method are calculated analytically, resulting in high accuracy and computational efficiency. In addition, an element agnostic approach has been developed leveraging primary analysis matrices to calculate their derivative. This characteristic, along with the nonintrusivity, makes the method employable with standard commercial software. Both approaches have been applied and validated in a wide range of scenarios, involving vibration and buckling problems.

On the Complementarity between CMMN and iStar in Complex Domain Modeling

Case Management Modeling and Notation (CMMN) and iStar are two distinct, multi-purposed modeling techniques that may be used to represent organizational challenges at separate levels of abstraction. CMMN, a flexible process-oriented technique, aims to extract knowledge that enhances the representational capacity of activity flows for a specific case. Conversely, the iStar framework adopts a goal-oriented modeling approach, effectively capturing the interplay of social actors and their influence on the attainment of organizational objectives. While prior studies have explored methods for integrating these techniques to attain a more comprehensive understanding of the organizational landscape, these remain primarily associated with a distinct level of the organizational matrix (i.e., CMMN for operations and tactics, and iStar for more strategic aspects). As such, any effort to evaluate their semantic proximity appears fragmented, as it deals only with the partial association of specific notations and elements. This article describes the conduct of a dual-purposed literature review to identify specific criteria that might accommodate a more holistic assessment of the two modeling techniques; these criteria are then employed as a guiding framework to construct a set of propositions that articulate the areas of complementarity and/or divergence between these two techniques, as identified in previous research. These propositions are subsequently subjected to validation by domain experts, leveraging a real-world case study in the educational domain. The results show that there can be areas of semantic convergence between the two techniques, suggesting their parallel use to effectively model complex domain problems. Overall, the present study aims to crystallize an approach for conducting complex modeling comparisons that transcends technical considerations.

Solving Poisson Problems in Polygonal Domains with Singularity Enriched Physics Informed Neural Networks

Solving Poisson Problems in Polygonal Domains with Singularity Enriched Physics Informed Neural Networks

Meshless method for wave propagation in poroelastic transversely isotropic half‐space with the use of perfectly matched layer

AbstractNumerical investigation of wave propagation in transversely isotropic poroelastic half‐space with the use of a new stretched coordinate system through the Meshless Local Petrov–Galerkin (MLPG) formulation is presented in this paper. To this end, the u−p formulation of Biot is adopted as the framework of the porous media. One approach to numerically solve the infinite domain problems is the use of an absorber layer in which the whole half‐space is divided into two parts, that is (i) a finite part, in which the responses are interested, and (ii) the remaining semi‐infinite part, which is replaced by a Perfectly Matched Layer (PML). The stretched coordinates in the PML are introduced in such a way that the wave propagating in it does not generate spurious reflection to the finite part. Comparing the numerical results with some existing exact solutions and evaluating the norm of error demonstrate that the response functions in the finite part are achievable as precise as desired. Some new results are also presented which show the validity of the numerical approach in poroelastic transversely isotropic domain.

Data-driven simulation-assisted-Physics learned AI (DPAI) for heat diffusion in large grain polycrystalline materials

In this paper, we propose Data-driven simulation-assisted Physics-learned Artificial Intelligence (DPAI), a deep-learning algorithm to simulate heat diffusion in large-grain polycrystalline materials. The DPAI model captures the spatio-temporal representation of heat diffusion in the material from input sequences from the training dataset. The training dataset consists of various temperature plots of polycrystalline materials taken from Finite Element (FE) simulations having varying numbers of grains oriented in random directions with a single-point heat source at the center. The arbitrary plane of the 3D microstructure of these materials is represented using 2D Voronoi tessellations. Voronoi configurations are used to model the geometry of the 2D Computer-Aided Design (CAD) model. Each cell of the Voronoi tessellation represents one grain of the microstructure. This CAD model is used as an input to the FE for solving heat diffusion equations. To model the near-realistic material anisotropy and accurately measure temperature differences at cell boundaries, a smaller mesh size is required in FE modeling, which takes considerable solver time. Therefore, the proposed Deep learning model significantly reduces the computational time while maintaining accuracy as compared to conventional numerical techniques. After training, the effectiveness of the trained DPAI model is examined by modeling larger domain problems involving a greater number of grains and varying material properties. The simulation result is qualitatively compared with the experiment. A scaled-up version of the microstructure is represented using Unidirectional Carbon Fiber laminate. The laminate is heated with a point heat source and the temperature plots are captured using Infrared Camera.

A new method for solving fractional optimal control problems

In this article, we present a novel approach that utilizes the operational matrix method to solve fractional optimal control problems. We address the challenge of missing initial conditions for the co-state function by incorporating the linear shooting method. Our objective is to provide a method that not only enhances computational efficiency and reduces cost but also improves accuracy and usability. With our new approach, we can calculate the coefficients of the expansion solution without the need to solve an algebraic system. These coefficients can be generated explicitly or through an iterative process. We provide a proof of the uniform convergence of the approximate series of functions to the unique solution of the optimal control problem. To demonstrate the effectiveness of our proposed method, we solve several numerical examples and compare the results with those obtained by other researchers. Additionally, we extend the application of this new method to solve problems with multiple states, thereby expanding its scope to a wider range of problem domains. This allows us to address diverse scenarios using the same efficient and accurate approach. Through our research, we contribute to the development of a powerful and versatile method for solving fractional optimal control problems, offering significant advantages over existing techniques.

Operator estimates for problems in domains with singular curving of boundary

Operator estimates for problems in domains with singular curving of boundary

Exponential Convergence and Computational Efficiency of BURA-SD Method for Fractional Diffusion Equations in Polygons

In this paper, we develop a new Best Uniform Rational Approximation-Semi-Discrete (BURA-SD) method taking into account the singularities of the solution of fractional diffusion problems in polygonal domains. The complementary capabilities of the exponential convergence rate of BURA-SD and the hp FEM are explored with the aim of maximizing the overall performance. A challenge here is the emerging singularly perturbed diffusion–reaction equations. The main contributions of this paper include asymptotically accurate error estimates, ending with sufficient conditions to balance errors of different origins, thereby guaranteeing the high computational efficiency of the method.

A multi-agent learning framework for mixed-integer linear programming

Mixed integer linear programming (MILP) is an important problem in the combinatorial optimization domain, which has wide applications in practical optimization scenarios. Given that most MILP problems fall into the NP-hard category, which the traditional methods may fail to solve, recent research has tried to derive MILP solutions using machine learning techniques. The whole MILP-solving procedure involves lots of modules, such as pre-solving, cut selection, node section, etc., and these modules are closely related and influence each other. However, the previous machine learning-based approaches neglect the connections between these modules, and focus on single-module learning techniques. To address this, we propose an initial step towards a more comprehensive multi-agent learning framework that allows different modules to interact and collaborate. Specifically, our current implementation involves two key modules: HEM for cut selection applied at the root node and GCNN for variable selection. By employing HEM to influence the training of GCNN, these two agents thus work in unison. Through extensive experiments on four MILP datasets in diverse scenarios, we observe significant improvements in solving time and PD integral metrics compared with the state-of-the-art learning-based MILP solving methods. This work lays the groundwork for future development of a fully integrated multi-agent framework.

Harmonic Mean Optimizer (HMO) for global problems solving

This study introduces the Harmonic Mean Optimizer (HMO), a novel meta-heuristic algorithm designed to address complex global optimization problems. The primary objective of this research is to develop an optimization technique that effectively balances exploration and exploitation without requiring extensive parameter tuning, thereby simplifying the optimization process and enhancing its robustness across various problem domains. The HMO employs a unique dual-fitness index that leverages the harmonic mean to assess both the quality and diversity of candidate solutions dynamically. This approach ensures a balanced search process that avoids premature convergence and improves the ability to find global optima. The methodology involves extensive benchmarking of the HMO against a comprehensive set of test functions, including 23 traditional functions from the Congress on Evolutionary Computation (CEC) 2017 test suite. The performance of the HMO is compared with that of established meta-heuristic algorithms, such as Genetic Algorithms (GA) and Particle Swarm Optimization (PSO), to validate its efficacy in terms of convergence speed and solution accuracy. Additionally, the HMO is applied to several real-world engineering problems to demonstrate its practical utility. The results show that the HMO consistently outperforms the benchmark algorithms, achieving faster convergence and higher accuracy in finding optimal solutions across a diverse range of test problems. In practical applications, the HMO effectively optimized complex engineering tasks, achieving significant improvements in both solution quality and computational efficiency. These findings highlight the potential of the HMO as a versatile and powerful tool for global optimization tasks. the HMO offers a significant advancement in optimization methodologies by providing a robust, parameter-free approach that effectively balances exploration and exploitation. This study underscores the HMO’s applicability to a wide range of optimization challenges and sets the foundation for future research and development in enhancing and applying this innovative algorithm to more complex and diverse problem domains.

Symmetrical fully coupled numerical model for efficient dam–reservoir interaction analysis in time domain

ABSTRACT A numerical model is presented for solving the dam–reservoir interaction (DRI) problem in the time domain. The primary aim of this paper is to introduce a coupling method that could be simply implemented in a monolithic DRI model, yields a symmetrical system of equations, and maintains accuracy comparable to existing studies. The finite element method is employed to solve the equation of motion for the structure, and the finite volume method is for the Navier–Stokes equations in the fluid domain. These two solvers are coupled using an innovative approach to generate a pressure-based symmetric system of equations, which is resolved implicitly. Moreover, unlike most previous DRI studies, free surface waves are considered in the modeling procedure, and their effects are investigated. Comparison of the results with benchmark test cases from the literature demonstrates that the proposed numerical method is both straightforward and precise.

Local and global existence for the Ericksen - Leslie problem in unbounded domains

The work deals with the Ericksen-Leslie System for nematic liquid crystals on the space RN with N≥3, on R+3 and on Ω⊆R3 exterior domain with sufficiently smooth boundary. The crystal orientation is described by unit vector v that is a small perturbation of a fixed constant vector η. We prove through a combination of energy method with dispersive a priori estimates a local existence and global existence for small initial data by a contraction argument. In particular, we obtain the following regularity of the liquid velocity u and of the crystal orientation vu∈L∞((0,T);Hs(Ω)),∇u∈L2((0,T);Hs(Ω)),∇v∈L∞((0,T);Hs(Ω)),∇2v∈L2((0,T);Hs(Ω)) for s>N2−1 if Ω=RN and s∈(12,1] if Ω=R+3 or in the exterior case, asking low regularity assumptions on u0 and v0.

Efficiently solving large-scale electrostatic lightning problems by integrating characteristic basis functions into boundary element method

The evaluation of the electric field near objects of complex geometries is critical in the pre-bridging regime of lightning strikes in high-voltage transmission systems, which is subject to an open domain problem and can be computationally expensive. In this paper, the characteristic basis function method (CBFM) is incorporated into the conventional boundary element method (BEM) to efficiently handle large-scale electrostatic problems. Within the CBFM, low-level basis functions have been replaced by high-level macro characteristic basis functions, resulting in a significant reduction in the number of degrees of freedom. This reduction in unknowns allows iteration-free direct solvers to approximate the matrices, further reducing the computational complexity of the problem. Numerical examples of the surface charge distribution of an actual 500 kV tower-transmission-line system and a three MW wind turbine in the lightning environment demonstrate the accuracy of the proposed CBFM, while requiring considerably much fewer computational resources compared with conventional BEM. Furthermore, CBFM is highly parallelizable, implying the feasibility of handling large electrostatic lightning problems with complex geometries.