This paper presents a novel approach to intrusion detection by integrating traditional signature-based methods with the contextual understanding capabilities of the GPT-2 Large Language Model (LLM). As cyber threats become increasingly sophisticated, particularly in distributed, heterogeneous, and resource-constrained environments such as those enabled by the Internet of Things (IoT), the need for dynamic and adaptive Intrusion Detection Systems (IDSs) becomes increasingly urgent. While traditional methods remain effective for detecting known threats, they often fail to recognize new and evolving attack patterns. In contrast, GPT-2 excels at processing unstructured data and identifying complex semantic relationships, making it well-suited to uncovering subtle, zero-day attack vectors. We propose a hybrid IDS framework that merges the robustness of signature-based techniques with the adaptability of GPT-2-driven semantic analysis. Experimental evaluations on a representative intrusion dataset demonstrate that our model enhances detection accuracy by 6.3%, reduces false positives by 9.0%, and maintains near real-time responsiveness. These results affirm the potential of language model integration to build intelligent, scalable, and resilient cybersecurity defences suited for modern connected environments.
Critical infrastructure systems, including energy grids, healthcare facilities, transportation networks, and water distribution systems, are pivotal to societal stability and economic resilience. However, the increasing interconnectivity of these systems exposes them to various cyber threats, including ransomware, Denial-of-Service (DoS) attacks, and Advanced Persistent Threats (APTs). This paper examines cybersecurity vulnerabilities in critical infrastructure, highlighting the threat landscape, attack vectors, and the role of Artificial Intelligence (AI) in mitigating these risks. We propose a hybrid AI-driven cybersecurity framework to enhance real-time vulnerability detection, threat modelling, and automated remediation. This study also addresses the complexities of adversarial AI, regulatory compliance, and integration. Our findings provide actionable insights to strengthen the security and resilience of critical infrastructure systems against emerging cyber threats.
With the rapid evolution of artificial intelligence (AI), face forgery videos have proliferated, posing significant societal challenges. Traditional detection methods struggle with poor generalization and cross-database accuracy, unable to address subtle features and variations in face images across scales and compression levels. This paper reviews current face forgery detection methods, identifying key limitations. It introduces a novel model enhancing features through knowledge distillation, optimizing generalization and robustness via a unique loss function and temperature adjustment strategy. Additionally, a Discrete Cosine Transform with multi-scale and multi-compression capabilities (DCTMS) is integrated, enriching texture and detail capture. Experimental results on deepfake datasets demonstrate the efficacy of the proposed methods, achieving high detection accuracy and robustness across diverse scenarios, including cross-database experiments. This study contributes valuable insights and techniques to advance the field of face forgery detection, addressing risks associated with manipulated video content.
Knowledge graphs (KGs) are typically used to enhance the performance of recommender systems by leveraging their characteristic to supplement the sparse user-item interaction data in the latter. This is because KGs feature abundant semantic information about entities (e.g. users or items in recommender systems) and inter-entity relationships. Deep learning has been seen gain wide use in boosting the performance of recommender systems. Thanks to focusing merely on nodes in a KG, or entities, such an approach, however, suffers from well capturing the semantics hidden in the node connectivity when learning node embeddings, leading to undesirable discovery of potential semantic associations between nodes. Against this background, we propose for recommender systems a multi-level cross-view contrastive learning mechanism that learns high-quality feature representations from unlabeled data. By incorporating item-item correlations into item embeddings, we developed an interitem semantic association graph (SAG), which was intended to model more comprehensive and finer-grained inter-item semantic associations. A series of experiments have been carried out and the findings verify that the proposed mechanism outperforms other baseline models on all metrics, confirming the effectiveness of coupling SAGs with contrastive learning in enhancing the performance of recommender systems.
Geographic clustering of businesses holds significant importance in understanding local economic dynamics, identifying areas of commercial activity, and assisting in spatial analysis for economic development. Artificial intelligence (AI) driven analysis is employed in this paper to investigate patterns of geographic clustering, particularly focusing on legal businesses within a given area. Data extraction techniques help preprocess business directories and classification codes to aggregate business addresses and visualize their spatial distribution. Clustering algorithms are used in conjunction with Geographic Information System (GIS) tools for data visualization and precise mapping, with respect to economic indicators. Expected outcomes include generating geographical distribution maps, comparing clustering algorithm results, and insight into urban business clustering patterns. This research considers potential external factors influencing business agglomeration and data currency. Recommendations focus on integrating AI-driven analysis with GIS tools and future research domains. Overall, this paper highlights the intersection of AI and geospatial analysis, providing stakeholders with valuable insights into the spatial distribution of economic activities within a target area.
There is a very serious problem plaguing people right now, and that is the shortage of blood banks around the world. This is because blood is at the heart of a major healthcare challenge, blood plays a vital role as the body’s energy source. To address this issue, this paper describes a post-donation blood quality testing system, as well as a cross-matching testing system for blood type suitability for patients, that utilizes the advanced security features of Amazon Web Services (AWS), rigorously stores all user data, including personal information and blood type, and adds a commitment to protecting the integrity and confidentiality of blood donor data. AWS’s scalability ensures that the system can adapt to the growing demand for blood supply while maintaining performance and security, enhancing trust and transparency between donors, recipients, and healthcare providers. Blood is at the heart of a major healthcare challenge. Blood plays a vital role as the body’s energy source. Hence, we propose a web-based application system that is integrated in real-time with hospital databases in various regions, especially in remote areas, enabling rapid identification of blood needs and effective distribution, greatly reducing emergency response time. Also, the system will effectively help people living in remote areas. This system not only solves today’s blood supply management challenges but also sets a new standard for global healthcare equity and access in the future, ensuring that life-saving blood is available everywhere and to everyone who needs it.
In the realm of private permissioned blockchain platforms, increasing throughput is a pivotal objective. This paper focuses on the optimization of throughput in Hyperledger Fabric, a private permissioned leading blockchain framework tailored for enterprise applications. The paper proposes a novel approach to enhancing the platform’s performance by reevaluating the endorsement policy. By implementing a Less-Than-Half endorsement policy, the paper aims to streamline transaction validation processes and bridge the gap between Fabric’s throughput and the demands of large-scale industrial applications. The proposed method objects to boost transaction throughput without compromising security or reliability. The paper provides an overview of Hyperledger Fabric architecture, discusses the pre-verification mechanism, and presents the proposed method for optimizing throughput. The performance of the system is analyzed using Hyperledger Caliper and Prometheus. Simulation results show increase in the throughput of the Less-Than-Half of the endorsement policy as compared to the majority and it also demonstrates the significant reduction in the latency of the Less-Than-Half endorsement policy.
The 5G mobile network aims to enhance wireless communication by providing faster and reliable connectivity. Open radio access network (RAN) architecture, which offers flexibility and innovation in Radio Resource Allocation (RRA), is central for optimal network performance. However, traditional RRA methods fall short of meeting the complex demands of 5G due to scalability issues. Incorporating machine learning (ML) techniques into open RAN can enhance adaptability and intelligence, ensuring that 5G networks meet high performance and service quality standards. This paper presents a comprehensive review of ML-based and traditional RRA methods in meeting the evolving demands of wireless networks. Literature from relevant articles were selected and analysed to highlight the techniques used, trends, strengths, and limitations. The findings reveal the potential and transformative impact of ML on the future of wireless communications, particularly in achieving the key performance indicators and quality of service expected from 5G and beyond networks. It also shows that research in ML-based RRA methods is at its infancy stage and more research is needed to advance the technology.
In recent years, machine learning has been widely used to process large-scale data and train complex models. However, there are certain security risks in the construction process of machine learning models. Data trainers may infringe on the privacy of data owners and use personal data to train machine learning models without authorization. To truly prevent unauthorized training of personal data, this paper proposes a privacy-preserving personal data sharing system, called PPDS. And this paper adopts the combination of CP-ABE with outsourced decryption and secure multi-party computation to ensure the secure authorization and training of personal privacy data. In PPDS, data users can only obtain the trained models instead of the authorized data, effectively preventing the privacy leakage of personal data. Moreover, the experimental results show that PPDS is a safe and effective scheme to achieve personal data sharing.
This paper presents a new blockchain scheme using a linkable ring signature algorithm based on lattice cryptography. This scheme counters quantum attacks on blockchain transaction signatures. The signature algorithm ensures correctness, anonymity, and unforgeability using trap generation and rejection sampling. It’s resilient to quantum attacks due to lattice cryptography’s hardness. Compared to other lattice-based algorithms, it offers reduced signature generation and verification time, minimized signature length, lower storage requirements, and better scalability. Implementing this scheme supports blockchain transaction security and privacy, promoting blockchain’s sustainable development and widespread application. This research enhances blockchain security against quantum attacks, paving the way for more secure and efficient blockchain-based apps.
In a highly competitive market environment, companies utilize intelligent media technology and animation design elements to build brand image, attract consumer attention, and influence their preferences. Animated brand logos, with their eye-catching appeal, can significantly impact consumer responses. This paper aims to explore the influence of visual brand elements, such as animated logos, on consumer purchasing behavior. A total of 233 participants were involved in this study, with 103 exposed to static logos and 120 to animated logos. The study results indicate that both static and animated logos affect consumer purchase intentions. By designing and implementing experiments with animated brand logos, this research helps to broadly understand the role of visual elements in consumer decision-making and provides specific recommendations for enhancing communication effectiveness. Additionally, it emphasizes that intelligent media, through context-aware computing, analyzes consumer behavior and preferences to deliver tailored content, thereby enhancing user experience.
This paper describes the essential role of Knowledge Management (KM) in transforming Moroccan public administration, underlining the fusion of ethical leadership with the push for digital modernization. It showcases KM as a pivotal tool in dismantling bureaucratic barriers and facilitating digital transitions, thereby streamlining public service delivery. By analyzing existing KM frameworks relevant to the public sector, this research formulates a customized set of implementation guidelines tailored to the Moroccan context. These guidelines strategically integrate ethical governance with KM practices to overcome the unique challenges faced by the Moroccan public sector, aiming to enhance service efficiency, transparency, and responsiveness. The study progresses from reviewing significant KM literature to spotlighting key factors for successful KMS deployment, culminating in specific, actionable recommendations. This concise exploration contributes to the discourse on KM’s capacity to revolutionize public administration, providing a blueprint for future research and practical applications within Morocco and potentially other similar contexts.
This paper introduces the Adaptive Multi-Layered Non-Terrestrial Network (AMLT-NTN), an architecture that integrates satellite, High Altitude Platform Stations (HAPS), and Unmanned Aerial Vehicles (UAVs). It leverages a combination of Free-Space Optical (FSO) and Radio Frequency (RF) communications, tailored for specific operational altitudes to enhance connectivity in remote and disaster-stricken regions. The AMLT-NTN tackles the complexities of dynamic power allocation and link selection by incorporating real-time optimization algorithms. This significantly boosted the network’s robustness and adaptability to environmental challenges and demand fluctuations. Simulations in OMNeT++ highlighted a quantifiable enhancement, with up to a 30% increase in throughput and a 40% decrease in latency, outstripping conventional NTN. The AMLT-NTN architecture demonstrates unparalleled resilience, consistently delivering high service levels across various conditions. Looking ahead, this research paves the way for integrating emerging communication technologies and scaling the architecture for widespread adoption. The proposed AMLT-NTN offers transformative solutions for rural connectivity and rapid disaster response, thus poised to impact global digital inclusion.
A large number of active bots in social networks pose a significant threat to the security of these networks. Although models based on Graph Transformer have achieved good results in bot detection tasks, they tend to focus more on global information, thereby neglecting the local information from neighbor nodes that are more similar to the target node. This phenomenon is known as over-globalization. We propose DGBot, a contrastive learning model based on cluster partitioning and relational graph convolutional networks (RGCN), which injects more local information into the Transformers model, thereby improving the model’s ability to extract local information. Extensive experiments on two widely-used public datasets demonstrate that DGBot outperforms competitive baselines.
With the advancement of society and economy, projects are becoming larger and more complex, requiring project managers to have a higher level of expertise. The emergence of big data technology presents both opportunities and challenges for project management. By effectively managing a variety of information, project delivery times can be reduced, risks can be mitigated in a scientific manner, and project schedules can be optimized. This article explores the use of big data in project management, focusing on schedule control and optimization analysis, offering fresh perspectives for innovation in project management. In today’s complex project environment, it is essential to fully leverage technology to improve project management skills and overcome the challenges posed by the abundance of data.
Signed social networks, characterized by positive and negative attributes on their edges, offer a nuanced view of relationships between users. However, most existing datasets are outdated and fail to capture current social dynamics. Additionally, as an important type, neutral links are not considered in most datasets. Hence, this paper introduces a novel signed social network dataset based on recent comments from YouTube videos in 2024. Unlike most traditional datasets, our dataset includes three types of relationships—positive, negative, and neutral—providing a more comprehensive representation of user interactions. Thus, our dataset can provide a more updated and diverse data for signed network research.
This paper presents our efforts in building a framework of automatic sentiment analysis for online user product reviews. The framework consists of web scraping to acquire a corpus of publicly available user product reviews, construction of a reference dictionary for removal of noisy entries such as typos and special characters from the reviews, extraction of Bagof-Words (BoW) representations, building BoW-based sentiment models for cosine similarity classification, neural networkbased classification, and performance evaluation and linguistic analysis modules. In a case study on Amazon user reviews, we analyze the impact of polarity salience- and exponential-based weighting of unigram and bigram BoW vectors and salience weight-driven dimensionality reduction on sentiment prediction accuracy and show that both strategies provide a dramatic reduction of equal error rates. Finally, we use the salience weighting matrix established in the classification experiments to rank-order discriminability of dictionary entries and their bigrams and report our findings in the form of word clouds. The dataset collected for this study, as well as the experimental framework, are made publicly available.
Large Language Models (LLMs) are widely used in natural language processing tasks due to their powerful semantic understanding and knowledge integration capabilities. Numerous existing recommendation studies consider recommendation tasks as a type of natural language processing, and thus LLMs have consequently brought new changes to the recommendation system paradigm. Existing research on recommendations using LLMs partly utilizes their rich data information, fine-grained user profiling, and expanded recommendation content to improve recommendation effectiveness. Additionally, some and partly studies directly uses LLMs to implement a generative recommendation paradigm. This paper adopts the literature review method to systematically sort out the current research status of news recommendation based on LLMs and classifies and summarizes the relevant research. To comprehensively understand the research in the field of news recommendation using LLMs, this paper introduces the current major work in the field of news recommendation from the two categories of generative LLM-assisted recommendation and direct generative recommendation and summarizes the current work as well as the potential future research directions and challenges.
Depth information in images is crucial for applications in autonomous driving, 3D reconstruction, and robot navigation. Self-supervised depth estimation methods have gained considerable attention because they rely solely on unlabeled video inputs. Traditional sensing equipment, such as LiDAR, incurs high costs and has limited effectiveness in adverse driving conditions, making image-based depth estimation a more economical and versatile alternative. However, real-world applications often involve diverse and noisy environmental conditions, whereas most existing models are trained on clean, ideal datasets. Our paper presents an advanced deep-learning method for robust self-supervised depth estimation, specifically designed to operate effectively under various dynamic scenes and noise conditions, which is called DSG. Our approach takes into account the noise and disturbances that can occur in 3 driving scenarios. Results on standard datasets demonstrate the model’s effectiveness, showing excellent robustness in handling noisy and diverse conditions. These strategies enhance the decision-making and safety evaluation capabilities of autonomous vehicles, facilitate the widespread adoption of low-cost sensing devices, and significantly improve model robustness in variable environments. The proposed self-supervised depth estimation method substantially enhances the reliability and accuracy of deep learning models, especially in dynamically changing application scenes.
Under specific environmental conditions such as dense fog or high dust levels, conventional RGB imaging technology faces significant challenges in capturing clear images. In contrast, infrared imaging technology, due to its unique characteristics, can effectively acquire images under these adverse conditions. However, the high cost associated with improving image quality through hardware enhancements in infrared imaging makes software-based image quality improvement crucial. Recent studies have demonstrated that deep learning networks hold significant potential for enhancing the quality of super-resolution images. To address the issues of gradient vanishing, insufficient feature utilization, and feature redundancy present in deep learning networks, this paper proposes a dual-channel hybrid convolutional residual network based on CNN with super-resolution of infrared images, which combines dual-feature extraction and dense linking. The network employs channel splitting to effectively reduce feature redundancy and leverages residual and mixed convolution techniques to enhance feature extraction and utilization. This approach efficiently preserves image details while eliminating noise. Comparative analysis using qualitative and quantitative metrics demonstrates the effectiveness of the proposed network for infrared image super-resolution tasks. The effectiveness of the network proposed in this paper in the task of super-resolution of infrared images is demonstrated by comparing it with other methods in terms of qualitative and quantitative metrics.
