Journal of Artificial Intelligence, Applications and Innovations

This study was conducted with the purpose of “developing and validating a model of the convergence of artificial intelligence and content marketing in the digital transformation of businesses.” The research design followed a sequential exploratory mixed-method approach. In the qualitative phase, semi structured interviews were conducted with experts in communication, digital marketing, and artificial intelligence. The data were analyzed using the thematic network approach at three levels—components, dimensions, and core concepts. The output of this phase yielded five principal concepts: “content creation,” “use of automation tools,” “efficiency of data analysis,” “effective interaction and content distribution,” and “enhancement of decision making,” which formed the basis for developing the quantitative instrument. The content validity of the items was confirmed through expert judgment and the Content Validity Ratio (CVR) with a threshold of 0.62. In the quantitative phase, the final questionnaire was distributed online, and 489 valid responses were collected. A confirmatory factor analysis (CFA) was performed; nine items with low factor loadings, as well as the “search engine optimization” factor, were removed to improve model fit. The measurement model fit indices and chi square ratio were found to be satisfactory. The final outcome presents a coherent and reliable framework that clarifies the linkage between the technical capacities of artificial intelligence and the content strategic needs of startups, offering a practical roadmap for designing and assessing content quality, implementing automation, optimizing distribution, and strengthening data driven decision making.

This research investigates the efficiency of Artificial Intelligence (AI) techniques in forecasting cyberattack trends and contrasts their performance with traditional approaches in environments compliant with ISO/IEC 27001. Using simulation-based evaluations, various AI algorithms including Neural Networks, Random Forests, Support Vector Machines, and Bayesian Networks were tested alongside conventional threat detection methods such as signature-based detection and heuristic analysis. The results demonstrate that AI-driven methods surpass traditional ones in several critical metrics. Neural Networks achieved the highest detection accuracy of 97.0% and delivered the fastest incident response times at 1.2 seconds, outperforming traditional techniques that exhibited lower accuracy and slower response. Additionally, AI-based anomaly detection models like Isolation Forests successfully identified emerging attack patterns with superior detection rates and faster processing times. Bayesian Network models also provided enhanced risk assessments, aligning more closely with ISO/IEC 27001 compliance requirements compared to classical methods. Although AI solutions involve higher upfront costs, they offer improved cost-effectiveness and overall performance over the long term. This study highlights the significant benefits of integrating AI into cybersecurity frameworks, emphasizing its role in advancing threat detection, response efficiency, risk management, and regulatory adherence.

Reviewing scientific articles and comparing their status can identify scientific gaps and potential opportunities. This study focuses on the field of hybrid models of banking and artificial intelligence (AI). AI applications in banking have grown significantly, ranging from fraud detection and risk assessment to personalized customer services and automated trading systems. These technologies are not only enhancing operational efficiency but also transforming how financial institutions interact with their customers and manage risks. In this paper, after extracting data from the Scopus database, categorization was performed on 4,795 reputable articles over the past 14 years (2010-2023). Clusters were created using text mining techniques to assign subject labels in the interdisciplinary fields of AI and banking. The Box-Jenkins approach was then used to select a model on the data and predict and analyze trends over different periods. The results indicate the primary focus areas for applying AI in banking are: Innovation, Technologies and Digital Banking (58.89%), Commercial and Investment Banking (27.13%), Retail, Personal and Wealth Management Banking (9.49%), and International and Global Operations Banking (4.48%).

In this study, we present an enhanced U-Net-based model for effective image denoising, incorporating a hybrid attention mechanism that combines both spatial and channel attention. These dual attention blocks enable the network to dynamically focus on relevant features while suppressing noise across both dimensions, thereby improving denoising performance. To further refine the output and enhance perceptual quality, a Gaussian filter is applied as a post-processing step, resulting in smoother edges and better texture continuity. The model also leverages Batch Normalization and Dropout techniques to stabilize training and prevent overfitting. Experimental evaluations were conducted on the CIFAR-10 and DIV2K datasets using standard performance metrics. The proposed model achieved an accuracy of 82%, a loss of 0.01, a PSNR of 37 dB, and an SSIM of 0.94—outperforming several state-of-the-art denoising methods. These results confirm the model’s strong ability to preserve structural and textural image details while significantly reducing noise. The combination of convolutional deep learning, hybrid attention mechanisms, and post-processing filtering offers a powerful and scalable solution for image restoration tasks. Furthermore, it demonstrates strong potential for practical applications in real-world scenarios such as image quality enhancement and medical imaging.

Tuberculosis remains a major global health challenge, and the World Health Organization has endorsed artificial intelligence tools to support imaging-based screening. This umbrella review mapped AI applications across tuberculosis prevention, diagnosis, and treatment, focusing on data modalities, algorithm types, and validation practices. Following PRISMA guidelines, systematic reviews published between January 2020 and June 2025 were identified through PubMed and Web of Science, and eligible studies were screened, extracted, and appraised in duplicate. Twelve reviews covering 648 primary studies were included from an initial 1,796 records. Accuracy, AUC, sensitivity, and specificity were reported in 54 %, 43 %, 45 %, and 40 % of studies, respectively. Only 33 % conducted internal validation and 4 % performed external validation against independent cohorts or human readers. Deep learning models such as VGG16, ResNet50, and InceptionV3 dominated, while classical algorithms like support vector machines and random forests persisted in contexts requiring interpretability or when data were limited. The findings highlight rapid growth in AI use for tuberculosis but reveal inconsistent reporting, limited external validation, and minimal attention to missing data. Broader clinical adoption will require rigorous, multimodal studies with standardized performance metrics, transparent algorithms, and robust validation strategies.

The evolution of language models marks one of the most transformative trajectories in the history of Natural Language Processing (NLP). This survey aims to provide a structured overview of key developments, tracing the progression from early statistical models to deep learning approaches, and culminating in the rise of Transformer-based architectures and Large Language Models (LLMs). We categorize and synthesize key contributions based on algorithmic paradigms, performance metrics, and systemic challenges. Specifically, we examine contributions from foundational models such as n-gram and Hidden Markov Models (HMMs), advances enabled by Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks, and the paradigm shift introduced by self-attention mechanisms in Transformer architectures. Additionally, the survey discusses how LLMs have expanded the capabilities of NLP systems in tasks including text generation, translation, and dialogue modeling. Alongside these achievements, we critically highlight ongoing challenges, including model bias, interpretability, computational costs, and environmental impacts, drawing on recent literature and evaluation frameworks. Emerging trends toward improving model efficiency, fairness, and societal alignment are also explored. By mapping historical progress and identifying open questions, this article offers a comprehensive reference for researchers and practitioners interested in the evolving landscape of language models.

Federated Learning is a distributed machine learning paradigm designed to preserve user privacy on decentralized devices without transferring raw data to a central server. Protecting data privacy in FL involves determining permissible operations and how they can be executed. This review provides an in-depth exploration of privacy threat models within FL, distinguishing between scenarios where the central server is either trusted or untrusted, and identifying appropriate defensive tools and technologies for these settings. The review covers secure computational techniques, including MPC, HE, and TEEs, as well as privacy-preserving mechanisms such as DP, LDP, and DDP models. It also examines hybrid approaches that combine multiple privacy models to enhance efficiency and robustness. The effectiveness of these methods is analysed across different scenarios involving both honest and potentially malicious servers and users. The findings reveal that while privacy-preserving methods mitigate risks, challenges persist in trade off privacy, communication efficiency, and model accuracy. This review highlights open research directions and serves as a comprehensive reference for researchers and practitioners seeking to implement robust privacy measures in federated learning systems.

Security anomaly detection, a critical element in safeguarding digital systems, has undergone a transformative evolution through the integration of deep learning techniques. This comprehensive review navigates the landscape of security anomaly detection, unveiling the potential and challenges within this realm. The research methodology involved systematic data collection from renowned databases, including Scopus, Web of Science, and Google Scholar. Key topics explored include the integration of deep learning models, benchmark datasets, preprocessing techniques, ethical considerations, and future directions. Deep learning models, such as autoencoders, recurrent neural networks (RNNs), and convolutional neural networks (CNNs), have proven invaluable in enhancing detection accuracy and efficiency. Benchmark datasets like NSL-KDD, CICIDS2017, and UNSW-NB15 have emerged as essential evaluation tools. Tailored preprocessing techniques ensure data readiness for these models. Challenges encompass data imbalance, model interpretability, adversarial attacks, and scalability. Ethical and privacy considerations emphasize privacy preservation, fairness, transparency, and accountability. The convergence of deep learning with security anomaly detection heralds a new era in cybersecurity. While challenges persist, a commitment to ethical principles and exploration of innovative avenues are set to realize the full potential of deep learning for robust, efficient, and responsible security anomaly detection systems, ensuring a safer digital landscape for all.