Optimization techniques have received significant attention for reliably addressing practical problems. A potential meta-heuristic called elk herd optimizer (EHO) was created, inspired by the social behavior and reproduction of elks. EHO has drawbacks, including poor convergence competency and a tendency to fall into local extrema in various optimization problems. Furthermore, this algorithm does not account for the memory of its search agents and has difficulty effectively balancing exploration and exploitation, which can lead to early convergence toward a local optimum. This study addresses the above issues by proposing an ameliorated EHO (AEHO) by incorporating several modifications into the basic EHO algorithm, which can be described as follows: A new hybrid memory-based EHO is developed that uses the particle swarm optimization (PSO) algorithm to guide EHO to search for reasonable candidate solutions. This hybrid approach was proposed to enhance EHO’s diversity and balance search capabilities to achieve strong search performance. Initially, a memory component was added to EHO using the idea of pbest from PSO to tap into promising search regions, which focuses on improving the best solutions and preventing the algorithm from getting stuck in a local optimum. In addition, the PSO concepts of (gbest) and (pbest) are used to enhance the best placements of the search agents in EHO. Finally, a greedy selection method was used to improve the efficiency of exhaustive exploration in AEHO, using the fitness values before and after updates as an indicator for efficacy of the best solutions. To evaluate the performance of the AEHO algorithm against a group of well-known competitors, we use ten complex test functions from the global CEC2022 test suite and thirty complex test functions from the global CEC2014 test suite. Based on the analysis of the experimental findings, AEHO performed optimally on 84% of the CEC2014 functions and 74% of the CEC2022 functions, ranking first in both suites with an average ranking of 3.11 and 1.62, respectively. The mean computation time of AEHO is about one-third of the average computation time for the first-ranked method, indicating that AEHO not only performs very well in global searches but also exhibits greater search efficiency when compared to newer optimization algorithms. The applicability and reliability of AEHO were thoroughly studied on four constrained engineering design problems and a real-world industrial process. The results demonstrate the superiority and promising potential of AEHO in addressing a wide range of challenging real-world problems.

Accurate prediction of rod-pumping system efficiency is crucial for evaluating the performance of such systems. Currently, the efficiency of rod-pumping systems is primarily estimated using mechanistic models. With the continuous advancement of information technology and the improvement of oilfield databases, some researchers have employed single neural networks for prediction. However, single neural networks often suffer from low prediction accuracy and poor robustness to noise. To solve this problem, we propose a new integrated learning-based soft measurement of the efficiency of rod pumping systems. Firstly, we proposed five soft measurement methods for rod pumping system efficiency: BiGRU-BiLSTM-CrossAttention, BiRNN-BiGRU-KAN, CNN-BiGRU-KAN, BiLSTM-BiGRU-KAN, and BiLSTM-Transformer-KAN. Then, using these five methods as base learners and FNN as the meta-learner, we constructed a novel rod pumping system efficiency soft measurement method based on the Stacking ensemble learning framework. The hyperparameters were optimized using a multi-strategy integrated Crayfish optimization algorithm, and the model was validated using 5-fold cross-validation. To verify the accuracy of the proposed soft measurement method, we applied it to 10,250 real oil wells for calculation and conducted a comparative analysis with baseline models. The results demonstrate that the proposed soft measurement method can effectively predict the efficiency of rod pumping systems.

Abstract Application of Artificial Intelligence (AI) in healthcare is growing exponentially, and its use is expected to continue expanding in the coming years. However, lack of trust in AI systems remains a significant barrier to their widespread adoption. This article analyzes the problem of trust, its various features and its application in AI-mediated healthcare. We first review the literature on trust and trust in technology to detect which theoretical constructs are essential to trust. We then identify the factors that we consider fundamental for a rich and complex comprehension of trust in AI-mediated healthcare. We finally propose a trust-factor framework that could be used for empirical research on AI-mediated healthcare and its practical implementation.

Abstract Machine Learning (ML) has revolutionized various domains, offering predictive capabilities in several areas. However, there is growing evidence in the literature that ML approaches are not always used appropriately, leading to incorrect and sometimes overly optimistic results. One reason for this inappropriate use of ML may be the increasing availability of machine learning tools, leading to what we call the “push the button” approach. While this approach provides convenience, it raises concerns about the reliability of outcomes, leading to challenges such as incorrect performance evaluation. In particular, this paper addresses a critical issue in ML, known as data leakage, where unintended information contaminates the training data, impacting model performance evaluation. Indeed, crucial steps in ML pipeline can be inadvertently overlooked, leading to optimistic performance estimates that may not hold in real-world scenarios. The discrepancy between evaluated and actual performance on new data is a significant concern. In particular, this paper categorizes data leakage in ML, discussing how certain conditions can propagate through the ML approach workflow. Furthermore, it explores the connection between data leakage and the specific task being addressed, investigates its occurrence in Transfer Learning framework, and compares standard inductive ML with transductive ML paradigms. The conclusion summarizes key findings, emphasizing the importance of addressing data leakage for robust and reliable ML applications considering tasks and generalization goals.

Artificial intelligence (AI), machine learning (ML) and decision-support (DS) are gaining increasing interest with widening adoption. This article investigates the enabler role of AI and ML for providing decision-support in procurement&purchasing domain. The study follows a systematic review approach via taxonomic analysis. Comprehensive analysis and discussions are provided for: (a) the relevance and applicability of AI and ML in procurement&purchasing decision-support; (b) functionalities/processes for which they are utilized; (c) related methodologies; and (d) implementation benefits as well as challenges. Findings reveal that procurement&purchasing area holds significant potential in terms of AI-ML applications for decision-support almost every related sub-process. This study is original by offering a process-oriented approach to the research domain; providing unique clustering and classification; and presenting detailed analyses via unique taxonomy tables with respect to approach, topic, focus, context and methodologies of the literature items reviewed. The study offers further research opportunities and has significant potential to provide managerial insights by the identified sectoral applications, benefits and challenges.

Abstract This paper presents a comprehensive survey of the applications, challenges, and limitations of Generative AI (GenAI) in enhancing threat intelligence within cybersecurity, supported by real-world case studies. We examine a wide range of data sources in Cyber Threat Intelligence (CTI), including security reports, blogs, social media, network traffic, malware samples, dark web data, and threat intelligence platforms (TIPs). This survey provides a full reference for integrating GenAI into CTI. We discuss various GenAI models such as Large Language Models (LLMs) and Deep Generative Models (DGMs) like Variational Autoencoders (VAEs), Generative Adversarial Networks (GANs), and Diffusion Models, explaining their roles in detecting and addressing complex cyber threats. The survey highlights key applications in areas such as malware detection, network traffic analysis, phishing detection, threat actor attribution, and social engineering defense. We also explore critical challenges in deploying GenAI, including data privacy, security concerns, and the need for interpretable and transparent models. As regulations like the European Commission’s AI Act emerge, ensuring trustworthy AI solutions is becoming more crucial. Real-world case studies, such as the impact of the WannaCry ransomware, the rise of deepfakes, and AI-driven social engineering, demonstrate both the potential and current limitations of GenAI in CTI. Our goal is to provide foundational insights and strategic direction for advancing GenAI’s role in future cybersecurity frameworks, emphasizing the importance of innovation, adaptability, and ongoing learning to enhance resilience against evolving cyber threats. Ultimately, this survey offers critical insights into how GenAI can shape the future of cybersecurity by addressing key challenges and providing actionable guidance for effective implementation.

Abstract LLarge language models (LLMs) demonstrate strong natural language processing capabilities but come with significant computational costs, raising questions about their practical utility compared to small language models (SLMs). This study systematically compares SLMs (DistilBERT, ELECTRA) and LLMs (Flan-T5, Flan-UL2) on two customer review analysis tasks: sentiment polarity classification and product correlation analysis. Our results show that while LLMs outperform in sentiment classification, they do so at a much higher computational cost, whereas fine-tuned SLMs excel in domain-specific correlation analysis with greater efficiency. To balance accuracy and cost, we propose a context-enhanced hybrid (CE-Hybrid) model, which refines traditional hybrid methods by enriching LLM input with SLM-generated insights, reducing redundant computation while maintaining accuracy. Our findings quantify the trade-offs between model performance and resource efficiency, offering actionable insights for businesses to optimize AI deployment. These results have significant implications for real-world applications such as e-commerce, customer service automation, and business analytics.

Cardiac imaging is vital for diagnosing coronary artery disease (CAD), with coronary computed tomography angiography (CCTA) being commonly used to evaluate coronary vessels for stenosis, calcification, and atherosclerosis. However, CCTA images often suffer from artifacts like beam hardening, scatter, and noise, degrading image quality and obscuring anatomical details, leading to diagnostic uncertainty. Conventional post-processing techniques, such as filtered back projection and iterative reconstruction, have limited effectiveness in correcting these artifacts, posing challenges in CCTA, where precise visualization of coronary arteries is crucial. Artifacts can blur vessel boundaries, obscure calcified plaques, and misrepresent stenosis severity, potentially leading to misdiagnosis and suboptimal clinical decisions. Recent advancements in computational imaging, particularly deep learning algorithms, offer clinical benefits for artifact reduction in CCTA. Deep learning models, such as convolutional neural networks (CNNs), outperform traditional methods by effectively de-noising and correcting artifacts through learning complex patterns from large datasets. These models adapt to the non-linear, heterogeneous nature of artifacts, enhancing image clarity and diagnostic reliability. Improved image quality in CCTA enables better visualization of coronary arteries, aiding in accurate assessment of stenosis and calcification. This review highlights deep learning approaches for artifact correction in CCTA, emphasizing their potential to improve CAD diagnosis.

Abstract Supervised machine learning recently gained growing importance in various fields of research. To train reliable models, data scientists need credible data, which is not always available. A particularly hard and widespread problem deteriorating the performance of methods are mislabeled samples (Northcutt in J Artif Intell Res 70:1373-1411, 2021). Common sources of mislabeling are weakly defined classes, labels that change their meaning, unsuitable annotators, or ambiguous guidelines for labeling. Because mislabeling lowers prediction quality, it is essential for scientists to be able to identify wrong labels before actually starting the learning process. For that, numerous algorithms for the identification of noisy instances have been developed. However, so far, a comprehensive empirical comparison of available methods has been missing. In this paper, we survey and benchmark methods for the identification of mislabeled samples in tabular data. We discuss the theoretical background of label noise and how it can lead to mislabeling, review categorizations of identification methods, and briefly introduce 34 specific approaches together with popular data sets. Finally, 20 selected methods are benchmarked using artificially blurred data with controllable mislabeling and a new real-life genomic dataset with known errors. We compare methods varying the amount and the type of noise, as well as the sample size and domain of data. We find that most of the methods have the highest performance on datasets with a noise level of around 20-30% where the best filters identify around 80% of the noisy instances with relatively high precision (0.58 $$-$$ 0.65). Acquiring precise predictions seems to be a more challenging task than identifying most of the noisy instances: while the average recall score over all models ranges from 0.48 to 0.77, the average precision score ranges from 0.16 to 0.55. Furthermore, none of the methods excels over all others in isolation, while ensemble-based methods often outperform individual models. We provide all data sets and analysis code to enable a better handling of mislabeled data and give recommendations on usage of noise filters depending on various dataset parameters.

This review explores the frontiers of large language models (LLMs) in psychological applications. Psychology has undergone several theoretical changes, and the current use of artificial intelligence (AI) and machine learning, particularly LLMs, promises to open up new research directions. We aim to provide a detailed exploration of how LLMs are transforming psychological research. We discuss the impact of LLMs across various branches of psychology—including cognitive and behavioral, clinical and counseling, educational and developmental, and social and cultural psychology—highlighting their ability to model patterns, cognition, and behavior similar to those observed in humans. Furthermore, we explore the ability of such models to generate coherent, contextually relevant text, offering innovative tools for literature reviews, hypothesis generation, experimental designs, experimental subjects, and data analysis in psychology. We emphasize the importance of addressing technical and ethical challenges, including data privacy, the ethics of using LLMs in psychological research, and the need for a deeper understanding of these models’ limitations. Researchers should use LLMs responsibly in psychological studies, adhering to ethical standards and considering the potential consequences of deploying these technologies in sensitive areas. Overall, this review provides a comprehensive overview of the current state of LLMs in psychology, exploring the potential benefits and challenges. We hope it can serve as a call to action for researchers to responsibly leverage LLMs’ advantages while addressing the associated risks.