Deep generative models are increasingly useful in medical image analysis to solve various issues, including class imbalance in classification tasks, motivating the development of multiple methods, where the Variational Autoencoder (VAE) is recognized as one of the most popular image generators. However, the utilization of convolutional layers in VAEs weakens their ability to model global context and long-range dependencies. This paper presents CAT-VAE, a hybrid approach based on VAE and Cross-Attention Transformers (CAT), in which a cross-attention mechanism is employed to promote long-range dependencies and improve the quality of the generated images. On the Ultrasound breast cancer dataset, the CAT-VAE achieved better image quality (FID 8.7659 for Malignant and 7.8761 for Normal) compared to the standard VAE. An experiment was conducted where a CNN classifier model was trained without data augmentation, with augmentation based on VAE, and using synthetic data generated by CAT-VAE. The CNN achieved the highest accuracy (97.00%) when trained with CAT-VAE synthetic images. A classification accuracy of 86.67% was achieved with mixed datasets of real and synthetic images, demonstrating that CAT-VAE improves generalization and resilience. These results highlight CAT-VAE's ability to produce diverse and realistic synthetic datasets.

Control charts are fundamental tools in Statistical Process Control (SPC), employed to monitor and improve production processes. This study aims to evaluate the performance of the Maximum Exponentially Weighted Moving Average (Max-EWMA) chart by incorporating Auxiliary Variables (AV) and Multiple Measurements (ME). This study also evaluating the covariate method, a multiple measurement framework, and scenarios involving linearly increasing variance. The evaluation focuses on the analysis of both Type I and Type II error rates, using simulation-based methodologies. The results indicate that the Multiple Measurement approach consistently outperforms the Covariate method, exhibiting lower Type II error rates and higher robustness in detecting process shifts. An increase in parameter A enhances the chart’s sensitivity to mean shifts, whereas parameter B shows negligible influence. Additionally, linearly increasing variance contributes to improved detection capability, particularly under conditions of high correlation. Overall, the Multiple Measurement method demonstrates strong effectiveness and reliability across a variety of conditions, underscoring its practical utility in process control applications.

In a previous work (Prog Addit Manuf 6:93–118, 2021), a novel Additive Manufacturing scan strategy was designed; the Skeleton-Based Perpendicular (SBP) scanning should show minimal trajectory series compared to classical exiting hatching patterns used in the literature. In contrast, this pattern should lead to mechanical anisotropy due to the one-way oriented printing if it is applied in all part’s layers; a complementary scan strategy must be designed to balance the SBP orientations. This important constraint led the author of this paper to develop the “Skeleton-Based Parallel” (SBPar) strategy as a SBP’s complementary scan for avoiding such issues. Subsequently, the present work details the design of the SBPar pattern and the corresponding scan length; analytical formulations are drawn-up for a simple rectangle as a proof of the concept. Therefore, the superposition of SBP and SBPar constitutes the total skeletal scanning (SB). Results emphasized two conflictual interests: apart from stripe scan, the proposed SBPar scan exhibits a maximized trajectory compared to the other scan strategies; thus, it seems lastly compromising the minimization objective targeted by SBP scan. On the other hand, according to this maximization aspect, the second interest is regarded in terms of surface control which requires maximizing matter spreading and thereafter offering higher densification to the processed surfaces. Furthermore, The SB and the classical scan strategies showed degrees of length-similarities according to decision variables adopted herein. Further works will be dedicated to the implementation of the Skeletal-Based trajectory within real 3D-parts and then to the associated mechanical characterization.

The main contribution of this paper center a round searching for a lost target which moves among finite number of cells according to truncated Markov chain with continuous time , the searcher distributes the effort among the states and the purpose is to minimize the search effort and the probability of undetection at the same time.

The conditional distribution of (X|θ) is regarded as the NXLindley distribution. This study is centered on the estimation of the Bayesian premium using the symmetric squared error loss function and the asymmetric Linex loss function, employing the extension of Jeffreys as non-informative priors and Gamma prior as informative priors. Owing to its complexity and lack of linearity, we rely on a numerical approximation for establishing the Bayesian premium. A simulation and comparison study with several sample sizes is presented.

This study tackles the energy management problem for wind distributed generators in AC microgrids (MGs) operating in both connected and isolated modes. A mathematical formulation is proposed to minimize energy losses and $CO$\(_2\) emissions, incorporating technical and regulatory constraints to reflect real-world MG operations. The solution methodology combines the Population-Based Genetic Algorithm (PGA) with an hourly power flow analysis based on the successive approximation (SA) method. To validate the proposed approach, a comprehensive comparison is conducted against three widely used metaheuristic algorithms: Particle Swarm Optimization (PSO), JAYA, and the Generalized Normal Distribution Optimizer (GNDO). Employing a rigorous statistical framework, including ANOVA and Tukey HSD tests, the algorithms' performance is evaluated through 100 independent runs per objective and configuration, using a 33-node AC microgrid with variable generation and demand as the test scenario. Results demonstrate that PGA consistently outperforms other algorithms, achieving lower mean values and variance in both energy loss and emission minimization. GNDO, by contrast, shows higher variability and less effective optimization. This work not only underscores the robustness and adaptability of PGA for sustainable microgrid management but also establishes a standardized framework for evaluating optimization algorithms in energy systems.

In the era of e-commerce, providing an exceptional customer experience is pivotal for online businesses. This paper introduces a comprehensive machine learning-based recommendation system meticulously crafted to enhance the customer experience on e-commerce platforms. Our system employs a multifaceted approach, incorporating product popularity analysis, model-based collaborative filtering, and textual clustering, to address a spectrum of user profiles and business contexts. It excels in delivering personalized product recommendations, effectively tackling the challenges associated with attracting and retaining new customers, as well as guiding businesses in their nascent stages of online presence. By harnessing diverse methodologies, this system not only optimizes the customer journey but also offers a versatile framework for future research endeavors aimed at continuously refining and adapting to the dynamic e-commerce landscape.

Partial least squares path modeling is a statistical method that facilitates examining intricate dependence relationships among various blocks of observed variables, each characterized by a latent variable. The computation of latent variable scores is a pivotal step in this method and it is accomplished through an iterative procedure. Within this paper, we investigate and tackle convergence challenges related to Hanafi-Wold's procedure in computing components for the PLS-PM algorithm. Hanafi-Wold's procedure, as well as alternative procedure, demonstrate the property of monotone convergence when mode B is considered for all blocks combined with centroid or factorial schemes. However, the absence of proof regarding the convergence of the error towards zero in Hanafi-Wold's procedure is a limitation compared to alternative procedure, which possesses this convergence property. Therefore, this paper aims to establish the convergence of the error towards zero in Hanafi-Wold’s procedure.

The subject of this article is IDS-Intrusion Detection Systems, which are strongly related to a comprehensive cyber attack prevention system. In the present day, an IDS for network infrastructure is a crucial topic. The advancement of SDN-Software Defined Networking has led to a rising need for software-based IDS-Intrusion Detection Systems. Diverse methodologies, including machine learning algorithms and other statistical models, have been used to develop distinct kinds of IDS-Intrusion Detection Systems to enhance performance. But still, that needs to be improved. Several studies have focused on solving these problems for this reason, utilizing methods like conventional machine learning models. However, existing systems need to improve, including low detection rate and high false alarm rate. The aim is to improve performance, specifically in terms of increases in detection rate. This work introduces a new IDS-Intrusion Detection System named SIDS-Stacked Intrusion Detection System, which utilizes a stack-based approach to improve detection accuracy and resilience. The objective is to utilize various predictive algorithms most efficiently. An ensemble classifier method is used to enhance the precision of the final prediction by amalgamating the outputs of multiple models. This research implemented numerous ML-machine learning methodologies, including Stochastic Gradient Descent, Logistic Regression, Random Forest, and Deep Neural Networks, to construct a multilayered model that would optimize network intrusion detection accuracy. This challenging research project employs the NSL-KDD dataset. In previous studies, the stacked model (DNN1 + DNN2) has a maximum accuracy of 97.90% for intrusion detection. However, the suggested trained model outperforms existing models by 98.40%. Additionally, the offered stacked model attains F1-score 99.2%, a FPR-false positive rate 95.6%, and a FNR-false negative rate 1.42%. In conclusion, the findings indicate that a stacked ensemble method can enhance evaluation metrics and provide more consistent performance.

A new compound extension of the Fréchet distribution is introduced and studied. Some of its properties including moments, incomplete moments, probability weighted moments, moment generating function, stress strength reliability model, residual life and reversed residual life functions are derived. The mean squared errors (MSEs) for some estimation methods including maximum likelihood estimation (MLE), Cram\'{e}r--von Mises (CVM) estimation, Bootstrapping (Boot.) estimation and Kolmogorov estimates (KE) method are used to estimate the unknown parameter via a simulation study. Two real applications are presented for comparing the estimation methods. Another two real applications are presented for comparing the competitive models. The nonparametric Hill estimator under the breaking stress of carbon fibers is estimated using the tail index (TIx) of the new model. Finally, a case study on reliability analysis of composite materials for aerospace applications is presented.