The rapid evolution of natural language processing technologies has significantly enhanced the capabilities of generative models, yet challenges remain in maintaining the accuracy and relevance of information over time. The novel concept of integrating a hierarchical Retrieval Augmented Generation (RAG) system with the Llama model addresses these challenges through enabling real-time access to vast external knowledge bases, thereby significantly improving contextual accuracy and relevance. The hierarchical RAG system, designed with multilevel caching and dynamic query routing mechanisms, optimizes information retrieval efficiency, reduces latency, and enhances the overall performance of the model. Comprehensive evaluations demonstrated that the RAG-enhanced Llama outperformed the baseline model across various metrics, including precision, recall, and F1-score, highlighting its superior capability in real-time information retrieval tasks. The system's adaptability to varying data loads and its efficient management of large-scale databases further underscored its robustness and scalability. Comparisons with traditional retrieval methods revealed the distinct advantages of the hierarchical RAG system, particularly in terms of retrieval speed and accuracy. These findings underscore the transformative potential of the proposed approach in advancing the field of natural language processing, providing a solid foundation for future research and development.

This paper addresses the claim that by enabling intelligent, flexible, and autonomous decision-making at the local scale, MAS can substantially strengthen the performance, reliability, and durability of smart grid operations. We discuss the design and implementation of a MAS for smart grid management through an in-depth case study, emphasizing on particular design decisions, difficulties faced, and results achieved. Among MAS's main benefits are its capacity to guarantee grid stability, reduce overload, and maximize the utilization of resources.

Artificial Intelligence (AI) and Cloud Computing (CC) are revolutionary technologies adopted by a wide variety of industries for managing and processing data. The integration of AI and CC enhanced the security of the cloud, better cloud resource allocation, scaling up data processing speed and size, intelligent data management, data analytics, data-driven decisionmaking, etc. On the other hand, CC provided a powerful computational platform to deploy complex AI tools with minimal hardware requirements. This synergy stood at the forefront of digital transformation post-COVID-19 era supporting businesses towards real-time predictive data analytics, and automated operations with advanced decision-making abilities. The synergy between AI and CC resulted in generative AI, one of the impactful innovations of the 21 st century which changed the way humans interacted with machines. This brief review paper attempts to highlight the recent advancements in various industries including Cloud Security, Healthcare, Semiconductor Chip Manufacturing, and Education.

Modern wireless networks must reliably support a wide array of connectivity demands, encompassing various user needs across diverse scenarios. Machine-Type Communication (mMTC) is pivotal in these networks, particularly given the challenges posed by massive connectivity and sporadic device activation patterns. Traditional grant-based random access (GB-RA) protocols face limitations due to constrained orthogonal preamble resources. In response, the adoption of grant-free random access (GF-RA) protocols offers a promising solution. This paper explores the application of supervised machine learning models to tackle activity detection issues in scenarios where nonorthogonal preamble design is considered. We introduce a datadriven algorithm specifically designed for user activity detection in Cell-Free Massive Multiple-Input Multiple-Output (CF-mMIMO) networks operating under GF-RA protocols. Additionally, this study presents a novel clustering strategy that simplifies and enhances activity detection accuracy, assesses the resilience of the algorithm to input perturbations, and investigates the effects of adopting floating-to-fixed-point conversion on algorithm performance. Simulations conducted adhere to 3GPP standards, ensuring accurate channel modeling, and employ a deep learning approach to boost the detection capabilities of mMTC GF-RA devices. The results are compelling: the algorithm achieves an exceptional 99% accuracy rate, confirming its efficacy in realworld applications.

Within few-shot learning, in-context learning (ICL) has become a potential method for leveraging contextual information to improve model performance on small amounts of data or in resource-constrained environments where training models on large datasets is prohibitive. However, the quality of the selected sample in a few shots severely limits the usefulness of ICL. The primary goal of this paper is to enhance the performance of evaluation metrics for in-context learning by selecting high-quality samples in few-shot learning scenarios. We employ the chi-square test to identify high-quality samples and compare the results with those obtained using low-quality samples. Our findings demonstrate that utilizing high-quality samples leads to improved performance with respect to all evaluated metrics.

This research investigates the viability of Nigerian barite deposits as a substitute for imported barite in drilling operations. Despite Nigeria’s abundant barite reserves, oil servicing companies have historically relied on foreign sources, hindering local production and economic growth. This study evaluates barite samples from three Nigerian locations - Nassarawa, Cross River, and Benue states - based on their chemical composition and physical properties, adhering to American Petroleum Institute (API) laboratory procedures. The results reveal that barite deposits from Calabar and Nassarawa meet the API standard, demonstrating their suitability for drilling applications. The findings suggest that processing Nigerian barite can generate significant revenue, equivalent to nearly one-third of the country’s 2023 budget. This underscores the potential for Nigeria’s barite deposits to contribute substantially to the nation’s economy. By exploring the utility of Nigerian barite as an efficient drilling fluid, this research aims to promote the development of Nigeria’s solid minerals sector, reduce reliance on imports, and support the growth of the oil and gas industry. The study provides valuable insights for policymakers, investors, and industry stakeholders, highlighting the importance of harnessing Nigeria’s natural resources to drive economic progress.

Ransomware poses a severe threat to cybersecurity, with its capacity to encrypt critical data and demand ransoms causing significant financial and operational damage to individuals and organizations alike. The hierarchical K-nearest neighbors (KNN) algorithm, proposed in this study, offers a novel and significant approach to ransomware detection by leveraging hierarchical relationships within opcode sequences to achieve more granular and precise classification. The methodology involves collecting opcode sequences from diverse ransomware and benign software samples, preprocessing the data through normalization, feature extraction, and dimensionality reduction, and applying a multi-level hierarchical KNN algorithm. Experimental results indicate that the hierarchical KNN algorithm outperforms traditional machine learning models in terms of detection accuracy, computational efficiency, and robustness to noise. The algorithm's scalability and moderate memory usage make it suitable for deployment in resource-constrained environments, highlighting its potential for practical applications in real-world cybersecurity scenarios. This research contributes a significant advancement in the field of ransomware detection, demonstrating the efficacy of hierarchical classification techniques in addressing the complexities of modern cyber threats.

Artificial Intelligence has enabled scientists to rapidly sift through and analyze massive collections of images, helping to identify objects worthy of closer study such as supernovae, pulsars, and quasars, and allowing us to classify stars, label galaxies, and other astronomical systems. Stellar classification serves as a cornerstone of astronomy, providing a framework for understanding and characterizing the diversity of celestial objects. This classification is based on the spectral properties. AI algorithms, specifically machine learning and deep learning models, have been designed to learn from data and make predictions or classifications based on patterns and trends. These models of spectral and stellar classification are trained on large datasets of known celestial objects, enabling them to recognize and categorize new objects based on their spectral signatures. Here we present a novel Fine Tunned Deep Convolutional Neural Network of 1D Separable Convolutional blocks for stellar classification based on spectral characteristics using SSDS-17 data captured by the Sloan Digital Sky Survey, where the class imbalance is assessed using the SMOTE balancing technique. The results obtained during the performance evaluation confirmed the reliability of the proposed architecture of StellarNet in multi-class stellar classification, obtaining remarkable values of around 97% and 99% for accuracy and AUC score respectively. The proposed StellarNet architecture can be applied to enable real-time classification of captured spectral characteristic data, facilitating automation of the task and providing labeled data for study and future research.

This paper presents a novel method for optimizing the outage performance of Reconfigurable Intelligent Surfaces (RIS)-assisted Rate-Splitting Multiple Access (RSMA) systems. The objective is to minimize the maximum outage probability for all messages to improve fairness. Instead of tackling the problem directly, which involves dealing with a non-convex nondifferentiable objective function, we propose a new formulation based on minimizing the Entropic Value-at-Risk (EVaR) of the perceived minimum rate among all messages in the system. We show that this formulation is equivalent to minimizing an upper bound on the maximum outage probability, then we transform it into a tractable smooth problem by devising a twice-differentiable approximation to it. In addition to its lower complexity, our method is shown by numerical results to outperform two benchmarks, namely, Discrete Exhaustive Search (DES) and Sequential Quadratic Programming (SQP)-based search methods.

Acute kidney injury (AKI) is a commonly encountered medical problem that is associated with poor health outcomes in AKI survivors, including increased mortality and re-admission to the hospital. Despite their high-risk status, only a small fraction (< 10%) of patients receive specialized nephrologist follow-up after AKI event. To address the gap in care for AKI patients, this work proposes an artificial intelligence (AI) based fusion technique that combines patient's single-lead electrocardiograph (ECG) and demographics to predict AKI recurrence 3-7 days before onset. The ECG data is analyzed with an on-chip reservoir-computer (RC) prototyped in 28nm CMOS process to create a compressed representation for predicting AKI onset from ECG. After fusion with demographics, the proposed technique is able to predict AKI recurrence 3-7 days before onset with 75.8% accuracy when evaluated on a retrospective patient dataset collected from Mayo Clinic Enterprise.

An Artificial General Intelligence (AGI) system must intelligently deal with the semantics of the situations it encounters. The flexibility required for intelligent responses is at odds with the fixed parameter spaces of statistical methods or the limitations of the rule-based syntax of the computing paradigms on which systems can now be built. Overcoming these limitations requires philosophical insights from the work of Kant, Wittgenstein, Heidegger, and Gadamer, covering ontology, epistemology, hermeneutics, axiology, and teleology. Specifically, additional teleological computations (to determine the purpose served, independently from the computations responsible for responding appropriately in the world), are necessary to overcome these shortcomings.These insights suggest using a new abstraction to satisfy the necessity for two computational processes to execute in parallel on their respective, related but different representations. The processes must consist of algorithms specifiable a priori, to apply in all possible situations. In this paper, we address some specific difficulties with concepts and show how these difficulties may be overcome using the new abstraction of higher-level perceptions proposed in the Subjective Emergence of Objective Minds (SEOM) model developed earlier.In this paper, we will focus on the representations and teleological computations necessary to give purpose to the actions of autonomous intelligent agents.

The article is devoted to the structure of the space industry of the Republic of Chile, its organisational and legal frameworks, as well as to the plans for its further development with a view to outlining promising areas for expanding Ukraine’s international cooperation with this country. The experience of the Republic of Chile, according to the author, is valuable due to its decentralised governance structure with regard to the space specialisation of each region, which is built on the use of both a favourable natural and geographical resources and the involvement of international projects.

The extraction of knowledge about the prevalent issues discussed on social media in Africa using Artificial Intelligence techniques is vital for informing public governance. The goals of our study are twofold: (a) to develop machine learning-based models to identify common topics of social concern about Africa on social media, and (b) to design a classifier capable of inferring a particular common topic associated with a given social media post. We designed a three-step framework to achieve the former goal, namely, topic identification. The first step uses text-based representation learning methods to generate text embeddings for feature representation. The second step leverages state-of-the-art Natural Language Processing models, commonly called topic modeling, to organize the representations into groups. The third step generates topics from each group, including the use of large language models to generate meaningful short-sentence labels from the bag-of-tokens associated with each group. To achieve the second goal of classification; we trained classifiers using ensemble voting and stacking learners to infer which among the identified common topics best characterizes the social media post. For our experimental study, we collected a text corpus called Social Media for Africa composed of 22,036 records extracted from social media comments on Twitter (X) and YouTube. The clustering-based model BERTopic yielded 304 topics, at topic coherence 0.81 C-v. On merging the topics into classes, the BERTopic+ created 11 common topic classes at topic coherence 0.76 C-v. We then utilized the identified topics based on the resulting groupings as labels for training a topic classifier. These labels were created using Llama2 on our SMA corpus. Our comparative study of topic classifiers using stacking and voting schemes shows that the BERTopic model features 0.83 accuracy and 0.82 F1 score with ensemble voting for training on topics. Furthermore, training on topic classes, BERTopic+ with ensemble voting had the highest accuracy of 0.95 and F1 score of 0.95 compared to other alternate methods on our corpus. The overall performance of classifiers using the ensemble stacking is slightly better than that of voting methods for short sentence topic labeling. For Africa, policymakers should focus on the most pressing social issues: COVID-19 restrictions affecting public health and economic recovery, promoting entrepreneurial innovation in energy and environmental sustainability to combat climate change, and strategically responding to China's rise in global politics to maintain geopolitical stability and foster international cooperation.

As the telecommunications industry embarks on the transition to Sixth-Generation (6G) networks, this paper examines the integration of Non-Terrestrial Networks (NTN), and in particular satellite backhauling, in the context of Fifth-Generation (5G) systems. The Integrated Access and Backhaul (IAB) technology, conceived as a wireless terrestrial backhauling system in the Next Generation Radio Access Network (NG-RAN), has been identified as a possible enabler for the integration of satellite nodes. Despite the work already done in this direction, the combination of IAB architectures with satellite nodes operating in both the access and backhaul side requires further evaluations on feasibility and limitations for networks integrating Low Earth Orbit (LEO) and Geostationary Earth Orbit (GEO) satellites. To this end, this work contributes providing insights on background technologies, as well as a detailed analysis of the issues and challenges arising from such integration and a definition of use cases to support narrow-band and broadband services. Furthermore, the design and implementation of a simulation tool is proposed for a performance evaluation in terms of registration time, link capacity, single-hop and end-to-end delay. Results show that the integration turns out to be feasible, even if with strong constraints coming from the satellite system rather than the IAB usage itself. Indeed, the earth-satellite link in LEO systems has a significant impact on the packet delivery time due to the discontinuous coverage. In case of GEO satellite instead, a non-terrestrial backhaul link could limit the performance of the whole system, especially at lower elevation angles.

3D fingerprint-based recognition and identification have several advantages compared to conventional 2D recognition systems. It is hygienic and secure due to its contactless sample collection. In addition to surface pattern, depth, curvature, and shape information can also be retrieved from 3D fingerprints which can be implemented for designing a more accurate and secure authentication system. While significant progress has been made recently, one of the great challenging issues is the fingerprint pixel’s topological height makes it difficult to extract fingerprint ridge/valley patterns. To address this issue, in this paper, the 3D fingerprint represented in 3D point cloud format is flattened using the B-spline curve fitting technique to reduce the impact of the pixel’s topological height. The flattened point cloud is converted to a gray-scale image by using the relative height of the points in the flattened 3D point cloud. The generated grayscale image is used for recognition via using a conventional 2D fingerprint identification method. The proposed method achieved an Equal Error Rate of 0.2974%, 0.28%, and 0.24% in three experiments, respectively, which is significantly more accurate than the existing methods.

Large Language Models (LLMs) have immense potential to transform the telecommunications industry. They could help professionals understand complex standards, generate code, and accelerate development. However, traditional LLMs struggle with the precision and source verification essential for telecom work. To address this, specialized LLM-based solutions tailored to telecommunication standards are needed. Retrieval-augmented generation (RAG) offers a way to create precise, fact-based answers. This paper proposes TelecomRAG, a framework for a Telecommunication Standards Assistant that provides accurate, detailed, and verifiable responses. Our implementation, using a knowledge base built from 3GPP Release 18 specification documents, demonstrates how this assistant surpasses generic LLMs, offering superior accuracy, technical depth, and verifiability, and thus significant value to the telecommunications field.

The advent of 6G wireless networks has the potential to unlock diverse applications of scalable autonomy. By advantageously coupling the individual and aggregated attributes of diverse multi-UAV fleets, a range of high-value applications such as logistics, enhanced disaster response, urban navigation, and surveillance can be significantly improved. However, enabling effective communication for knowledge fusion necessitates the intrinsic optimization of performance metrics like energy consumption, resource allocation, latency, and computational overheads to enhance autonomous efficiency. Furthermore, designing robust security features is essential to safeguarding privacy, control, and operational integrity. This paper explores a novel collaborative knowledge-sharing (KS) framework that leverages 6G and edge-computing capabilities to facilitate the cooperative training of decentralized machine learning models among multiple UAVs, without the need to transmit raw data. This framework aims to enhance the learning experience and operational efficiency of autonomous vehicles. The DECKS (distributed edge-based collaborative knowledge-sharing) architecture enables Federated Learning (FL) within UAV networks, allowing local models to be trained and shared among neighboring UAVs for creating global models. This promotes intelligent knowledge aggregation without a central entity, enhancing collaborative capabilities among autonomous vehicles. The DECKS architecture efficiently extracts and distributes collaborative shared experience to ground vehicles through edge and direct inference, reducing energy consumption, latency, and computational overhead. Our simulation analysis demonstrates that the DECKS architecture has the potential to reduce energy consumption by 70% in sensorless vehicles and improve autonomous vehicle learning performance by 15% compared to centralized approaches in a distributed environment. This improvement is achieved by comparing the efficiency of systems with and without aggregated knowledge.

Natural language processing models have shown lots of advancements in generating coherent and contextually relevant responses, yet they often struggle with retrieving precise and upto-date information due to the static nature of their training data. Introducing Federated Retrieval-Augmented Generation (RAG) represents a novel and significant approach by integrating federated learning with dynamic retrieval mechanisms to enhance information retrieval and response generation. This article presents the implementation of Federated RAG on Mistral 8x7b, an open-source large language model, demonstrating substantial improvements in retrieval quality and response accuracy. The federated learning framework facilitated distributed training across multiple nodes, ensuring data privacy while enabling the model to leverage diverse information sources. Comprehensive evaluation on the MMLU benchmark revealed that the Federated RAG model consistently outperformed the baseline RAG model, achieving higher accuracy and relevance in the generated responses. Detailed analysis and optimization of the retrieval mechanisms and training processes contributed to the model’s enhanced performance, highlighting the potential of Federated RAG as a scalable solution for knowledge-intensive applications.