Authors - Ezhilmathi Krishnasamy, Pascal Bouvry Abstract - The ongoing development in architecture and programming models, particularly regarding GPUs, significantly influences the landscape of high-performance computing. Presently, nearly all supercomputers worldwide are equipped with GPU compute nodes, promoting advancements in architecture that emphasize heterogeneity. Concurrently, there is a parallel evolution in programming models designed to effectively harness the potential of this diverse architecture for executing scientific applications. Notable architectures, longing trends in Nvidia GPUs, such as Grace Hopper and the AMD Mi300 series, alongside programming paradigms like SYCL, OpenCL, and various library-based models, including Kokos and StarPU, illustrate this trend. The primary motivation is to establish an appropriate programming model that targets modern architectures, essential for maximizing computational efficiency across specific applications. A multitude of programming models exist, each offering the potential for optimal utilization of advanced architectural capabilities. However, a critical question arises regarding the ease of adoption and the user-friendliness of these models for existing large-scale scientific codes. This study focuses on OpenMP Offloading, examining its application on Nvidia to ensure a unified source code compatible with various GPU architectures. Identifying an optimal programming model for effective OpenMP Offloading usage is of paramount importance. This paper conducts a comparative analysis of the OpenMP Offloading programming model against CUDA for Nvidia GPUs, facilitating a comprehensive performance evaluation. The analysis employs key BLAS operations to assess the performance characteristics of OpenMP Offloading in relation to CUDA, thereby elucidating the advantages and limitations associated with leveraging OpenMP Offloading.
Authors - Yury Zontov, Alexander Derendyaev, Daniil Tkachev, Arkady Vaynshtok, Vyacheslav Yakuba Abstract - A consulting system for development and adoption of high-quality decisions by territorial governments should provide tools for accessing relevant and reliable information about processes and results of population's life, as well as tools for analyzing socio-economic indicators and infrastructure functioning. Analysis tools should help identify problem situations ("bottlenecks") in ensuring the quality of life of population and to form a "tree" of management goals for the development of territory and improvement of life of population. The solution of these problems is possible by creating a human-machine platform for scenario analysis of the development of territories for numerous aspects of management. The decision-making process based on the representation, calculation and interpretation of current observed and predicted parameters is effective for both operational management and strategic planning. The information and analytical system under development is intended to become a convenient and powerful tool for analyzing the level of development of a region, scenario analysis, planning support and making high-quality management decisions.
Authors - Sunday Omotoye, Weifeng Chen Abstract - The escalation of insider threats poses significant harm to organizations and businesses, necessitating a thorough understanding of their nature and the associated threat landscape to develop effective Insider Threat Programs (ITP). This paper delves into the pervasive issue of insider threats within the retail industry, examining the applicability of the Cyber Kill Chain (CKC) to comprehend these threats. The retail sector, with its diverse and dispersed workforce, faces unique vulnerabilities that require organized strategies and proactive measures. By strategically integrating the CKC framework, the ITP systematically identifies and prevents insider threats at every stage of their lifecycle. This comprehensive approach serves as an indispensable element in fortifying the security framework of both the retail industry and broader economic landscape.
Authors - Sarmad K.D. Alkhafaji, Shahab Abdulla, Haydar Abdulameer Marhoon, Mohammed Diykh, Mustafa Ali Majed, Jafar Sadiq, Ali Aqeel Saleh, Aqeel Sahi, Hussein Alabdally Abstract - Chest radiography is a technique based on medical imaging that is employed to detect thoracic diseases. In this paper, we designed an intelligent method to diagnose thorax disease from chest X-ray (CXR) images. A novel empirical curvelet transform, coupled with a deep learning model, is proposed. The collected images are analysed using the proposed empirical curvelet transform (ECT) model. Then, the outputs of ECT model are sent to DenseNet. The proposed model is assessed using several statistical metrics. The proposed model achieves an accuracy of 98%. The results demonstrated the ability of the proposed model to detect Thoracic Disease.
Authors - Elias Tabane, Zenghui Wang, Ernest Mnkandla Abstract - In this paper, we present a comprehensive analysis of ensemble deep learning models for DNA sequence classification. We explore the performance of three standalone models: Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and Gated Recurrent Units (GRU), along with an ensemble approach that combines all three. Our study evaluates the models based on four performance metrics: accuracy, precision, recall, and F1 score. The ensemble model achieved an accuracy of 90.6%, with precision, recall, and F1 score all at 0.91. We compare these results to the standalone models and demonstrate that ensemble learning significantly improves classification performance in the context of DNA sequence data. Additionally, we review relevant studies that have applied deep learning models to similar tasks and discuss the advantages of combining CNN, BiLSTM, and GRU for sequence classification tasks.
Authors - Indibar Debnath, Om Prakash Abstract - For a prime p and ℓ ∈ N with gcd(ℓ, p) = 1, this work explores the hulls of ℓ-length cyclic codes over Zp3 . We establish the form of the generators of the hulls and provide a characterization of the hulls. Further, we present a condition for these codes to be LCD. To formulate the p-dimensions of the hulls, first, we find the types of the hulls and then use them to come up with an expression for the p-dimensions of the hulls. Moreover, for a fixed p-dimension of the hull, we count all the cyclic codes.
Authors - Shahriar Sultan Ramit, Nayeem Ahmed, Md Fatin Ishrak, Md Ruhul Amin, Alaya Parven Alo, Md. Sadekur Rahman Abstract - Oral Squamous Cell Carcinoma (OSCC) is among the most frequent cancer death causes, and early detection plays a vital role in improving patient survival. The traditional histopathological diagnosis is subjective and labor-intensive which necessitates an automated and standardized classification methods. This study has used a publicly available dataset comprising a total of 10,000 histopathological images. This paper evaluates four CNN architectures ResNet101, InceptionV3, MobileNetV2 and Xception to classify OSCC and normal cells. For better accuracy Hyperparameter Tuning was done on MobileNetV2. Tuned MobileNetV2 achieved the best performance with accuracy, recall, and F1-score of 0.99, demonstrating its efficacy in classifying malignant vs. normal tissues. To further enhance interpretability Explainable AI techniques were employed, including LIME and Saliency Maps, enabling visual comprehension of model predictions. Our results demonstrate the importance of deep learning for OSCC detection overcoming the "black-box" issue of CNNs by explain ability. This study contributes to AI-driven diagnostic innovation through a more accurate and interpretable approach to OSCC classification.
Authors - Nelson Salgado-Reyes, Jorge Rivera-Guaman Abstract - This proposal outlines the establishment of a Computer Security Incident Response Team (CSIRT) at the Central University of Ecuador (UCE) with a focus on training, awareness, and a hybrid model adapted to the academic environment. The study utilizes a mixed qualitative-quantitative approach to analyze international best practices, diagnose cybersecurity vulnerabilities at UCE, and design the proposed model. The findings emphasize the importance of strengthening security culture, reducing incidents caused by human error, and integrating cybersecurity into the university curriculum. The model includes an organizational structure, educational services, and assessment metrics that adhere to standards such as ISO/IEC 27035 and NIST SP 800-61. This research contributes to the field of CSIRTs in educational institutions with limited resources, providing a replicable framework specifically tailored for Latin American contexts.
Authors - Rabab A. Rasool, Muthana H. Hamd Abstract - This paper introduces a novel unimodal biometric system based on a derivative angle-based feature. Unlike traditional edge-based features, this approach leverages a one-to-one correlation between the angle and its corresponding edge, enabling robust and reliable recognition for a single-source trait biometric system. This unique property allows for performance comparable to multimodal systems, which typically rely on multi-source traits fusion techniques for improved accuracy. To evaluate the effectiveness of angle-based features, an extensive experiments is conducted on three diverse facial datasets (AR, FEI, and CK+) using 150 subjects. The facial features are extracted using seven distinct methods at varying levels, ensuring a comprehensive and fair comparison between edge and angle-based approaches. Recognition accuracy was assessed using various metrics, including False Acceptance Rate (FAR), False Rejection Rate (FRR), and three Error Distance Measures (EDMs): Euclidean, Manhattan, and Cosine distances. Results demonstrate that angle-based features achieve highly competitive performance compared to their edge-based counterparts. Across 210 recognition processes, edge-based features achieved an overall accuracy of 79%, while angle-based features recorded a closely comparable 73%. These findings highlight the potential of angle-based features as a promising approach for developing robust and reliable unimodal biometric systems.
Authors - Zakaria Kinda, Sadouanouan Malo Abstract - In Burkina Faso, farmers and plant pathologists remain with important challenges in treating cotton diseases. The development of natural language algorithms has advantages to the implementation of a tool for suggesting treatments for cotton diseases and the classification of cotton diseases using meteorological data. In order to propose a tool for suggesting appropriate treatments for diseases using large language models (LLM), this paper uses meteorological data collected by the National Agency of Meteorology of Burkina Faso (ANAM-BF) to 2014 and 2023 in addition to knowledge bases on cotton diseases. In this study, we compare the Llama2 model with the RAG system to the BERT model for classification and proposed of cotton disease treatments. The result of our approach obtains a 95.4% classification precision for cotton diseases with weather data. Users may interact with the tools to generate treatments for cotton diseases with the use of a console for Llama2 with RAG and a chatbot for BERT. The performance of Llama2 model with RAG to generate appropriate responses to cotton diseases in Burkina Faso was evaluated by comparing it with GPT.
Authors - Adarsh Varshney, Karthick Seshadri, Viswa Kiran Andraju Abstract - The Knowledge-Infused Policy Gradient with Upper Confidence Bound (KIPGUCB) strategy addresses contextual multi-armed bandit problems by balancing exploration and exploitation. This study evaluates the performance of a KIPGUCB-based agent in the partially observable environment of StarCraft II. Unlike traditional deep reinforcement learning models that rely on low-level atomic actions, our approach enhances decision-making by employing higher-level tactical strategies. A tactic manager dynamically selects optimal tactics based on game state and reward signals which improves resource management and structured tasks such as unit training. The agent’s performance is compared with a StarCraft II Grandmaster, a novice human player and DeepMind’s baseline RL agent across five mini-games. Experimental results show that the KIPGUCB-based agent outperforms the baseline model in resource focused and structured tasks but struggles in combat-oriented scenarios requiring adaptive responses.
Authors - Rajyashree H, Govindarajan J Abstract - The FMCW radar sensor technology has demonstrated remarkable effectiveness in various applications, both civilian and military, due to its affordability and efficiency in a wide range of operational conditions. Nonetheless, this technology faces challenges related to mutual interference during object detection, which can negatively impact its performance. Advanced signal processing techniques, combined with cutting-edge software tools, can help mitigate this issue. This study embarks on the innovative journey of developing a module within NS-3 a distinguished network simulator, that encompasses all essential components of FMCW radar, including transmission, reception, and processing units for the very first time. The prevailing challenges experienced by the researchers in asserting their protocols in real-time have motivated us to design and develop the radar module in NS-3 that fosters to validate the protocols for radars in heterogenous environment in an emulated setup. Two unique mitigation algorithms—Ti’sInterference Mitigation, and Frequency Hopping via Unslotted Aloha—have been meticulously implemented and assessed. This undertaking lays a firm groundwork for the enhancement of the radar module, facilitating better performance in intricate setups.
Authors - Vipin Bansal, Manisha Malhotra Abstract - Artificial Intelligence (AI) is transforming industries such as automotive, healthcare, insurance, and manufacturing through computer vision and Convolutional Neural Networks (CNNs) for image analysis. In medical imaging, AI enhances the interpretation of MRI, X-rays, and CT scans, reducing human error and improving diagnostic efficiency. Early detection of Diabetic Retinopathy (DR), a severe diabetes complication, is crucial to preventing vision loss. Traditionally, ophthalmologists manually analyzed retinal images to detect abnormalities like fluid leaks or lesions. AI now enables more precise and efficient analysis. This paper presents ViT-MADv2, an improved method for detecting DR-related abnormalities using Vision Transformer (ViT) generative models. As an extension of previous research [1], ViT-MADv2 enhances the base image generation module by incorporating diverse training data with variations in contrast, color, and lesion size. It also refines the similarity evaluator module to improve analysis. The model leverages a novel approach to compare embeddings from original and generated images, identifying DR-specific patterns. Experimental results demonstrate a 2% accuracy gain, reaching 96.5%, with improved sensitivity—a crucial factor in healthcare. These advancements strengthen AI-driven diagnostics, enhancing clinical confidence. Source code: https://github.com/vipinbansal1/vitmadv2.
Authors - Nuno J. P. Rodrigues, Tania S. S. Azevedo Abstract - The literature review on licensing and technology commercialization delves into the critical role that Intellectual Property (IP) management and effective licensing strategies play in the commercialization of technological innovations. Intellectual property, particularly patents, provides a competitive edge for firms, enabling them to appropriate financial returns from their innovations. Licensing emerges as a vital method for both leveraging technologies developed within organizations and universities and generating revenue streams. The review emphasizes the importance of robust IP management, which allows firms to secure their innovations and strategically negotiate licensing agreements to expand their market reach. The studies analyzed highlight the complex dynamics involved in technology transfer, with a particular focus on how universities play a pivotal role in transferring innovations to the market. University-based licensing officers often act as intermediaries between academic inventors and the commercial sector, facilitating the commercialization process. However, challenges such as bias in decision-making and the difficulties in securing timely licensing agreements, particularly in emerging fields like nanotechnology, are discussed. Additionally, the review explores how partnerships between companies and universities, through strategic alliances and well-structured contracts, can mitigate risks and enhance the successful commercialization of new technologies. Overall, this review synthesizes key insights into how companies and academic institutions can optimize their technology commercialization efforts through efficient IP management, strategic licensing, and collaborative partnerships.
Authors - VanVinh Le, HongGiang Nguyen Abstract - This study aims to use the 2025 Times Higher Education (THE) Asia University Rankings as a frame of reference to consider what drives and dictates university rankings in the ASEAN region. Their analysis looks at six major indicators – Teaching, Research, Citations, Industry Income, International Outlook and Rank.. Comparative analysis confirmed that while Vietnam universities are purely citation-dependent in their quest to enhance higher ranked tables, some universities of our Southeast Asian neighbors at least start from a more balanced and diverse evaluation. The study deployed three deep learning models BiLSTM, BiGRU, and a Hybrid BiGRU-BiLSTM to predict university rankings. Among all the models, the Hybrid model performed the best with the lowest RMSE (3.72) and MAE (2.61), which indicated the highest predictive accuracy and stability. The analysis additionally showed the power of higher citation scores to boost rankings, with the greatest gains accruing to institutions with the lowest ranking. Feature importance analysis indicated that collaboration between industry and research is very important for ASEAN universities. Quality of teaching was still a main focus in Vietnam universities. The results highlight the role of research impact and international collaboration to go up in ranking. The analysis cautioned that universities and colleges looking to secure their position on the world stage needed to prioritize the volume of research produced, degree of international engagement, and depth of industry collaboration. The findings guaranteed the suitability of deep learning models as a robust methodological toolbox for ranking prediction and data-driven strategic improvement of university performance.
Authors - HongGiang Nguyen, VanVinh Le, TrungKien Tran Abstract - This paper is an attempt to evaluate ASEAN universities’ research impact and more specifically that of Vietnam by analyzing research-oriented key performance indicators that drive rankings. According to data, these universities have large gaps between the average reputations for academia and employers, faculty-student ratios, and research output. The experiment's three models named LSTM, GRU, and Hybrid predicted rankings on citation performance. The results revealed that the Hybrid model produces the highest accuracy, reiterating that the citation per paper variable tends to most strongly drive the positive correlation between citations per paper and university ranking. In addition, the study also indicated that Hue University cited 175% more research opened up to gain below a rank of 55. This point emphasized that research output is a major driver of rankings and institutional prestige. At the same time, the study offered several recommendations for universities, particularly those looking to increase their rankings, including focusing on high-impact publications, developing global research partnerships, and investing in faculty development.
Authors - Abdullah I Alshoshan Abstract - Verification and/or identification (VI) of the individual writer-identity is one of the most common secure personal biometric authentications, particularly in banks for verification and in sensitive data storages for both VI. A writer-identity VI system (WIVIS) is proposed using the writer-invariant features of his/her script using two approaches: offline approaches, which rely on the script information in a static format, such as an image or shape, and online approaches, which require the collection of information in a dynamic format, such as speed and acceleration, using a tablet with a stylus pen to capture both of these dynamic information. Both offline script VI methods, such as normalized Fourier transform descriptor (NFTD) and normalized central moment (NCM), and online script VI methods, such as normalized script speed and acceleration, will be discussed. These features are compared individually and then as a combination. In the combination mode, the neural network (NN) is used for classification. Implementation and testing of the WIVIS is done and analyzed, and the effectiveness of each invariant algorithm, regardless of the language or form of the script shape, is discussed. A set of data on the online (dynamic) and offline (static) script is also discussed.
Authors - Dzung Lai Ngoc, Maria Luojus, Jukka Heikkonen, Rajeev Kanth Abstract - The development of cancer therapeutic therapies has made significant advancements in recent years. Numerous innovative solutions have emerged, achieving notable success, including immunotherapy, targeted drugs, and, among them, oncolytic viruses. Oncolytic virus therapy represents the first instance in which humans have employed a biological logic program, rather than a conventional drug, to treat a disease. Despite its promising potential, clinical trials involving oncolytic viruses have not yielded the anticipated outcomes, due to our incomplete understanding of the underlying biological logic and mechanisms. This paper will describe a treatment approach from a biological algorithmic standpoint, encompassing biological logic programs, molecules that carry biological logic (Logical Biomolecular Complexes - LBC), and the existing tools that can be used to design such treatment programs. Our proposal is based on a review of oncolytic virus studies, but the logic framework behind LBCs is tailored specifically for cancer treatment, rather than focusing on replication and spreading. This sets LBCs apart from their viral counterparts and can be considered a new concept.
Authors - Julian Andres Duarte Suarez, Leonardo Juan Ramirez Lopez Abstract - Given the continuous need of beds available for hospitalization in health institutions, and even more so during pandemic periods, it is necessary to have alternatives that allow patients to be transferred to their homes and from there to carry out continuous monitoring of their health. For this, DATALOG was developed, which is an Internet of Medical Things platform that acquires, processes, transmits, stores and manages the medical signals of patients from their home to a central hospital. Following this, the stored data is processed by applying the Standard Intersectoral Process methodology for data mining that al-lows them to provide the medical staff with the behavior of six physiological variables and visualize it in a unique control table developed in php. Initial tests show an acceptance of the medical staff as a service and support tool for medical decisions, and, in addition, it has been proven that hospitalization at home con-tributes significantly to the rapid improvement of the patient, thus decongesting the hospital system. Among the most recent innovations, DATALOG includes an early warning system that allows to warn about the patient's condition between normal, attention, alert and critical, with the possibility of sending the alert to mobile systems. It is concluded that DATALOG is a useful service and tool for e-health, and machine learning techniques allow for the prediction of patient states and alerts.
Authors - K P N S Dayarathne, U Thayasivam Abstract - Deep learning has achieved amazing success in multiple areas, such as Image Classification, speech recognition, Object Detection & Segmentation, natural language processing, audio-visual recognition, adaptive testing, etc., and is gaining major interest from the research community. The application for deep learning is growing day by day. Predicting interest rate as univariate analysis is important given that total spectrum of the interest rates is not available to apply yield curve analysis. This paper investigates the applicability of deep learning models such as RNN, LSTM, CNN and TCN to interest rates in Asina frontier countries such as Sri Lanka, Pakistan and Bangladesh. The deep learning approach for interest rate perdition is still under the radar, and this is the first attempt on the Asian Fronter market. Interest rates associates with Government securities were considered to have uniqueness for all three countries, where data range from 2010-2022 for Sri Lanka and Pakistan, whereas Bangladesh analysis was based on the same from 2015-2022. The results revealed CNN was the best model for Sri Lanka and Bangladesh, while LSTM was the best model for Pakistan based on the lowest RMSE. The study further investigates the applicability of different activation function for output layers and hidden layers, but found ReLU is the most viable activation function along with Max pooling. Further, it was found that CNN works better for countries with stable term structures of interest rates, and the immediate dynamics of the interest rate influence the near future interest rate.
Authors - Taeko Onodera, Koutaro Hachiya, Yuhei Hatakenaka Abstract - Numerous studies have applied machine learning to diagnosis and screening in the medical and welfare fields. However, it is rare for the resulting machine learning models to be widely adopted in clinical practice. This study proposes a method for deriving diagnostic rules from machine learning models that can be applied manually without the use of computers. The proposed method involves inputting all possible patterns into a trained model, generating a truth table with the corresponding prediction results, and then using the Quine– McCluskey method to derive logical expressions that serve as manual diagnostic rules. In the experiments, the proposed method was compared with conventional methods for deriving manual diagnostic rules from datasets: the point score system, a method based on likelihood ratios, and a logic derivation method based on rough set theory. Only the proposed method achieved a positive clinical utility index of 0.81 or higher—classified as “excellent”—even when the number of rules was limited to just two or three.
Authors - El Hadji Bassirou TOURE, Ibrahima FALL, Mandicou BA, Alassane BAH Abstract - Microservice architectures offer scalability and deployment benefits but introduce significant data consistency challenges due to distributed data ownership and the necessity of data duplication across services. Current approaches either compromise service autonomy with strong consistency mechanisms or shift complexity to developers. This paper presents Micro-Flex, a novel megamodel-based consistency management framework that treats data entities as component models with formally tracked consistency states.We extend the Modified-Shared-Invalid protocol with differentiated Shared states (Shared+ and Shared−) to accommodate varying consistency requirements while maintaining formal guarantees. Our approach formalizes Global Operation Models (GOMs) as application specifications where consistency states are actively managed through well-defined transition rules. We validate Micro-Flex through an e-commerce case study, demonstrating its effectiveness in balancing consistency guarantees with service autonomy while addressing data duplication challenges. 4 By applying Model-Driven Engineering principles to distributed consistency challenges, our framework contributes to more disciplined data management in microservice architectures.
Authors - Nikolaos Lazaropoulos, Ioannis Vaxevanakis, Ioannis Sigalas, Ioannis Lamprou, Vassilis Zissimopoulos Abstract - Cloud Native Functions (CNFs) support automated and dynamic orchestration of containerized network services, replacing traditional hardware-based architectures. These deployments consist of modular microservices that enable elastic scalability and collaborative service delivery. This paper presents an approximation framework for capacity constrained CNF resource allocation, modeled as variants of the Group Generalized Assignment Problem (Group GAP). The main contributions are: (1) a 1 2 -approximation algorithm for CNF placement when each function’s footprint is at most half the cluster capacity and (2) a 1 2 (1 − e−1/d)-approximation for shared microservices among multiple CNFs, where d is the degree of sharing, supported by experimental evaluation of the algorithm relative error.
Authors - Abdulaziz Alkhajeh, Sara Alhashmi, Alya Al Ali, Rakan Alhosani, Suhail Alshehhi, Deepa Pavithran, Joseph Anajemba Abstract - Healthcare has always been a crucial part of human life, with people investing resources to get the best services available. Ensuring patient confidentiality has always been crucial, but the digital era introduces new security risks. Hospitals now store patient information in computerized databases, which are vulnerable to cyberattacks. One major threat is ransomware attacks, where hackers capture sensitive and confidential patient data and demand large sums of money to prevent it from being leaked or sold. This puts patient privacy at risk and can disrupt healthcare services. Also, unauthorized access to the patients’ information compromising the data confidentiality has been a growing concern because health care has always been sensitive and personal information that should not be utilized for commercial purposes. Blockchain technology offers a solution by providing a secure way to store patient files. Using an Interplanetary File System (IPFS) on the blockchain, healthcare providers can save patient records in a decentralized and protected system, reducing the risks linked to traditional databases. This method helps protect patient information from cyber threats, ensuring privacy and security. In this paper, we are using blockchain-based architecture coupled with pinata IPFS cloud to secure the patient’s valuable information from any kind of cyber-attack, including ransomware.
Authors - Valinho Antonio, Eric Umuhoza, Pierre Bakunzibake, Moise Busogi Abstract - Accurate crop discrimination is vital for effective agricultural planning and sustainability management, especially in regions like Sub-Saharan Africa (SSA), where small-scale farming predominates and ground data is scarce. Conducting field surveys in SSA is challenging due to labor and cost constraints, as well as logistical and political barriers. This paper proposes a framework design of cost-effective satellite-based machine learning for crop type classification in crop growth with limited reference data. So, we have identified the important satellite timeseries features and the machine learning model architecture to be used to timely and accurately identify crops in a small and intercropped farms. This study therefore has great role on agricultural data collection at large scale which is one of the ways to accomplish food security advocated by the sustainable development goal two, zero hunger.
Authors - Asini Silva, Thushani Weerasinghe Abstract - The rapidly evolving IT job market presents a significant challenge with the growing gap between academic curricula and industry needs. Our study addressed this misalignment with a data-driven approach. The study employed a constructive research approach to develop a user-friendly platform that extracts real-time data from LinkedIn job postings and creates a dataset of IT jobs and their top-demanding skills. The authors utilized tools such as Selenium and BeautifulSoup to extract job titles and descriptions, identifying the most sought-after skills within the IT industry. Testing and evaluation were conducted weekly for several months, resulting in a dataset consisting of more than 50 IT job positions, each listing 20 in-demand skills. This comprehensive dataset not only captures but also retains records over time, allowing for an in-depth examination of skills associated with historical postings. The analysis uncovers notable discrepancies between the skills highlighted in academic programs and those required by employers, revealing a critical barrier to graduate employability. Hence, this research provides valuable insights for educational institutions, policymakers, and job seekers, equipping them to align programs with industry expectations while empowering students to make informed decisions regarding skill development. Furthermore, the methodology utilized in this study is scalable. It can be applied to any region in the world, offering a robust framework for aligning education with the evolving demands of the job market.
Authors - Jose Luis Chavez Torres, Tyrone Alexander Guarderas Cabrera, Camila Nickole Fernandez Morocho, KunYong Zhang Abstract - This article presents a comparative structural analysis of two housing typologies: one composed of reinforced concrete and the other of steel frame structures. The structural design was carried out using the force-based design method (DBF), considering the seismic and geotechnical parameters of the study area, obtained through SPT tests and bibliographic sources. The fundamental period of both structures was determined, and the corresponding response spectrum was generated, enabling the structural design and subsequent load descent ac-cording to the NEC-2015 combinations. In the foundation stage, three viable alternatives were proposed: isolated footing, combined footing, and mat foundation. Using the specialized software SAFE 2016, the performance of each option was evaluated based on soil pressure, punching shear resistance, and settlement behavior. Finally, a cost-benefit analysis was performed, considering concrete volumes and steel quantities, to select the most technically and economically suitable structural and foundation system for the study area.
Authors - Taufik Iqbal Ramdhani, Riri Fitri Sari Abstract - Real-time and accurate road infrastructure monitoring is a major challenge in urban areas. Traditional methods, such as manual inspections by municipal staff or vehicular surveys using costly technologies like LiDAR or laser scanners, are prohibitively expensive, geographically constrained, and deployed infrequently. To address this, crowdsourcing has emerged as an effective approach for expanding both the coverage and frequency of infrastructure monitoring. Building on this concept, CrowPotChain introduces a novel platform that combines AI-driven pothole detection with secure blockchain-based report submission, ensuring tamper-proof and reliable crowdsourced data collection. The framework utilizes the YOLOv11s-seg model for semantic segmentation, combining convolutional neural networks (CNN) with transformer-based elements, which provides impressive detection metrics (Precision: 0.889, Recall: 0.894, mAP@0.5: 0.944). Every verified report includes geolocation, date/time, and pothole size, securely embedded in a Proof-of-Work (PoW) blockchain for verifiability and immutability. To examine the system's performance, a benchmark was performed on four setups: no AI & no blockchain, AI only, blockchain only, and AI + blockchain, using batches of transactions from 10 to 100. The findings show that the no AI & no blockchain deployment provides the most rapid per-transaction time (approximately 0.010s), followed by AI only (0.059s to 0.135s), blockchain only (0.075s to 0.162s), and AI + blockchain (0.145s to 0.696s). Although blockchain does incur substantial overhead, its combination with AI can still serve response needs for civic infrastructure crowdsource reporting. Future development will add gamification, NFT, and IPFS to enhance participation, encourage reporting, and provide scalable decentralized storage.
Authors - Jiwan N. Dehankar, Virendra K. Sharma Abstract - The advanced incorporation of Machine Learning (ML) in blockchain systems present special challenges related to security, scalability, and adversarial robustness. The traditional consensus protocols and aggregation techniques suffer from high latencies, susceptibility to Byzantine node attacks, and inefficiencies in communicating gradients that cripple real-time federated learning on the blockchain networks. On the other hand, existing solutions like PoW (Proof-of-Work) and centralized aggregation do not adapt dynamically to ML workloads and remain vulnerable to adversarial attacks, thus putting the model's integrity into jeopardy and causing grave computational overhead. To mitigate these issues, we present Blockchain-Federated Secure Learning Network (BFSL-Net), an infrastructural framework with a dual purpose of enhancing security and efficacy while providing scalability to blockchain-based ML systems. BFSL-Net is comprised of (1) the Multi-Tiered Hierarchical Consensus Framework (MHC-BCML), (2) Adaptive Byzantine-Resilient Aggregation (ABRA), (3) Secure Adversarial Gradient Masking (SAGM-MLBC) and (4) Hierarchical Graph Neural Network-Based Threat Intelligence (HGNN-TI). BFSL-Net, which brings all the above-mentioned methods into a unified system of real-time threat resistance federated learning. Proposed model shows an adversarial threat mitigation success of 99.6 %, marked 2.8 times improvement in efficiency of ML processing, and a 4.5-fold reduction in blockchain computational overhead, thereby promising secure, scalable ML production environments in the blockchain space.
Authors - Natalya Aleynikova, Anna Loskutova, Mikhail Matveev, Elena Sviridova Abstract - The paper solves the problem of controlling the states of a system that may exhibit prolonged transient processes. As an example, the learning process of students is considered. Control is carried out based on predicting the dynamics of student academic performance. Academic performance – one of the key indicators of learning quality – is typically measured on graded scales and represented, for instance, by time series of student grades (scores). Instead of traditional time series analysis of grades, the present paper proposes transitioning to a space of fuzzy states (categories): “Fail”, “Satisfactory”, “Good”, and “Excellent”. The dynamics of these fuzzy categories are described using a discrete-time Markov chain model with fuzzy states, analyzing not the current but the limiting (steady-state) distributions of a student’s states. The paper presents a recurrent algorithm for the transition from the space of numerical grades to the space of fuzzy states, constructing the stochastic matrix of the Markov chain. The properties of the stochastic matrix are investigated to determine the existence and uniqueness of the limiting state distributions. Additionally, an approach is proposed for identifying change points – the moments when a shift in a student’s performance trend occurs.
Authors - Jose Luis Chavez Torres, Camila Nickole Fernandez Morocho, Tyrone Alexander Guarderas Cabrera, Ulbio Fernando Mendoza Hidalgo, KunYong Zhang Abstract - This article analyzes a detailed lithological characterization of the Punzara area, located in Loja, Ecuador. Through a systematic field survey and spatial analysis through a geological map at a scale of 1:2500, six different lithological units were identified and delimited. These include heterogeneous alluvial deposits, clayey deposits characterized by a remarkable and anomalous presence of organic roots at depths between 1.0 and 1.5 meters, a massive sandy silt unit with outcrops up to 3 meters thick, a conglomerate with spheroidal clasts of 1 to 4 cm in a silt-sandy matrix, another conglomerate with intercalated lenses of sand and clay, and shale outcrops with intercalations of gray silts with thicknesses of up to 30 meters. The spatial distribution of these units, clearly visualized through the geological map, provided fundamental geological information for the understanding of the local geological framework and lays the foundations for evaluating their potential influence on slope stability in the area.
Authors - Gusti Anisa Wulandari, Dewi Susita, Mohammad Sofwan Effendi Abstract - This study investigates the mediating role of Innovative Work Behaviour (IWB) in the relationship between Knowledge Sharing (KS), Work Engagement (WE), and Job Performance (JP) among state civil apparatus (ASN) at the Directorate of Road Transport, Ministry of Transportation of the Republic of Indonesia. Using a mixed-methods approach, the research employs Structural Equation Modeling (SEM) with AMOS to analyze survey data from 203 respondents, followed by qualitative exploration to contextualize the statistical findings. Results show that KS and WE significantly influence IWB, and although they also directly impact JP, IWB fully mediates these effects. Descriptive analysis reveals gender-based differences in IWB, with female respondents showing higher scores than males, while males reported slightly higher scores in KS, WE, and JP. These findings highlight the strategic importance of fostering IWB to optimize job performance. The study recommends organizational focus on strengthening IWB through targeted interventions, and suggests future research include gender and educational background as potential moderating variables, supported by qualitative methods to deepen understanding.
Authors - Syed Muhammad Raza Abidi, David C. Henshall,Gabriel-Miro Muntean Abstract - The weak electromagnetic signals originating from the brain’s neu-ronal activities can be assessed by electroencephalography (EEG) and magne-toencephalography (MEG). Due to the continuous time series data of EEG and low amplitude and nonstationary characteristics, it is difficult to achieve a con-sistent and satisfactory diagnosis outcome. It is hard to use these signals to identify and describe neuronal activation in the brain and adequate knowledge of signal processing, statistics, and numerical techniques are required. This paper introduces an innovative hybrid approach using machine learning, i.e., Modality Integration for Neuro Signals to Enhance Accuracy (MINE-Acc) which com-bines EEG and MEG data to increase brain activity prediction accuracy. This approach leverages the complementary strengths of both modalities to improve the accuracy and robustness of prediction. We employed the machine learning pipeline and used a Logistic Regression (LR) classifier in the research, performed a 5-fold cross-validation on sample dataset given by MNE-Python and by com-bining modalities together the findings provide a prediction accuracy of 99.8%. Traditional methods, such as functional magnetic resonance imaging (fMRI) and Positron Emission Tomography (PET) etc. are available to use these signals to characterize normal and pathological brain activity but there remain difficulties with integration and interpretation. They have high spatial resolution but lack real-time capabilities. The study determines the improved prediction accuracy of the activity participants engage with based on combined analysis of EEG and MEG data. We used the MNE-Python, a software package to test this novel approach.
Authors - Ahmed Noorim, Raina Nusrat Jahan, Md. Sabbir Al Ahsan, Sourav Adhikary, Md. Jamil Uddin Abstract - Early detection of Diabetic Retinopathy (DR) remains essential due to its status as a leading cause of vision loss along with severe complications. Ever increasing worldwide diabetes situation makes it necessary to develop an automated diagnosis system for detecting DR at an early stage precisely. The research proposes a weighted average ensemble deep learning architecture to perform two stages of DR diagnosis along with severity classification from fundus images. The first stage identifies cases of the presence of DR by applying a binary classifier followed by a multiclass classifier in the second stage to evaluate severity levels. This model is trained and evaluated on a merged dataset which amalgamates APTOS 2019, MESSIDOR 2, and IDRiD with three different preprocessing to boost its generalized application capabilities. DenseNet121, EfficientNetB1 and Xception models complement each other for optimal feature extraction and classification task thus utilized in the development of ensemble model. Outperforming several state-of-the-art models, 97% recall with 92% accuracy was achieved in determining the existence of DR, while severity classification reaches 93% accuracy and 94% recall. The research shows promising assistance for ophthalmologists in becoming an essential diagnostic screening instrument for the early detection of DR in resource limited areas.
Authors - Dario Galic, Dejan Stosovic, Elvir Cajic, Anita Katic Abstract - This paper explores numerical methods for solving partial differential equations (PDEs) using the method of nets. The focus is on hyperbolic equations, such as the wave equation, and the application of net methods in solving problems with boundary conditions. The process of solving these equations using computational tools is illustrated, and the accuracy of the results is analyzed. The iterative Gauss-Seidel method is applied to solve systems of algebraic equations generated by the net method.
Authors - Kamil Samara, Syed Rizwan Abstract - Mental health challenges, particularly anxiety and stress, are prevalent among students due to academic pressures, social expectations, and personal struggles. Traditional mental health support systems often fail to provide timely interventions, leading to severe consequences such as depression or suicidal ideation. This study presents an Early Detection and Support System for Student Mental Health, integrating machine learning models to proactively assess and predict student anxiety levels. The system utilizes data from activity logs, and survey responses to classify students into different anxiety categories and provide personalized support recommendations. Multiple machine learning algorithms, including Logistic Regression, Decision Tree, and Random Forest, were employed to optimize prediction accuracy. The results demonstrate that Logistic Regression achieved the highest accuracy (89.1%) in predicting stress levels, while the Random Forest model performed best in stress reduction prediction. The system's predictive capabilities extend beyond anxiety detection, enabling multi-feature mental health analysis, including depression, self-esteem, and stress levels. By integrating an automated alert mechanism and real-time monitoring, this framework offers a proactive solution for universities to support student mental well-being.
Authors - Alaya Parven Alo, Md Ruhul Amin, Md Imran Kabir Joy, Kazi Rezwana Alam, Shahriar Sultan Ramit, Md. Sadekur Rahman Abstract - Cervical cancer is a leading cause of cancerrelated deaths among women, and early detection is crucial for improving patient prognosis. Traditional diagnostic methods, while effective, are often timeconsuming and prone to subjectivity. This paper explores the use of deep learning techniques for automating cervical cancer diagnosis, employing five distinct models MobileNetV2, VGG19, Xception, ConvNeXtBase, and InceptionV3 along with a tuned version of MobileNetV2. A secondary dataset with five classes of cervical cell images were utilized to build the models and the performance of each model was evaluated with precision, recall and F1score. The tuned MobileNetV2 model achieved the highest accuracy and robustness in classification.TunedMobileNetV2 provided an accuracy of 0.99, with precision and recall values of 0.99.To address the "blackbox" nature of deep learning models, Explainable AI (XAI) techniques were incorporated, including LIME and Saliency Maps, to improve model interpretability. The use of XAI in the tuned MobileNetV2 model enhances transparency, allowing for visual interpretation of model predictions. Findings of the research suggests that deep learning, coupled with XAI, offers a promising and more explainable approach to cervical cancer diagnosis, advancing both accuracy and interpretability in automated clinical decisionmaking.
Authors - Katleho Seatlolo, Khutso Lebea Abstract - This paper investigates the potential of Retrieval Augmented Generation (RAG) technology to enhance the effectiveness of Intrusion Detection Systems (IDS) in the energy sector. By leveraging vast amounts of historical data, real-time threat intelligence, and advanced natural language processing techniques, RAG can significantly improve IDS capabilities in detecting and responding to cyber threats. The paper addresses the limitations of traditional IDS, such as their reliance on predefined signatures and vulnerability databases. It explores how RAG can overcome these limitations by analysing network traffic patterns, identifying anomalies, and correlating them with known attack vectors. The paper discusses the potential benefits of RAG in terms of improved threat detection accuracy, reduced false positives, and enhanced response times. Case studies and research findings are presented to support the argument. Challenges and considerations related to data quality, privacy, and ethical implications are also addressed. The conclusion emphasises the importance of RAG technology in safeguarding smart grids from evolving cyber threats and highlights potential future directions for research and development. The paper aims to explore the potential of RAG technology to enhance IDS in smart grids and contribute to Sustainable Development Goal 9: Industry, Innovation, and Infrastructure. It describes the limitations of traditional IDS, the benefits of RAG technology, and the potential applications of RAG in the energy sector.
Authors - Aditi Choudhary, Aditya Gupta, Pulkit Jain, Nikunj Agarwal, Mukund Wagh Abstract - We present a novel programming-by-example (PBE) approach that synthesizes natural and human-readable code by integrating higher-order functions with standard and third- party libraries in Haskell. This technique leverages refinement types to efficiently prune the search space, ensuring scalability while preserving soundness. Using Liquid Haskell, we extend support for complex data structures, enabling the synthesis of reusable and idiomatic code. Our evaluation demonstrates the tool’s versatility across lists, trees, maps, and domain-specific languages, including musical scores. The results highlight that our method generates concise, interpretable programs, bridging the gap between formal verification and practical usability in functional programming. Index Terms—Programming-by-Example (PBE), Haskell, Higher-Order Functions, Refinement Types, Liquid Haskell, Code Synthesis, Functional Programming, Search Space Pruning, Domain-Specific Languages (DSLs), Formal Verification, Scalability, Reusable Code, Idiomatic Programming.
Authors - Wanda Syauqi Mikola, Tiurida Lily Anita Abstract - This Research looks at how Customer’s Intentions to Buy from Fast-Food Restaurants are influenced by Online Reviews, Brand Images, and Menu Visuals. Understanding the factors that influence Customer’s Purchase Decisions is of business interest due to the impact of technology and the evolution of services. The objective is to empirically ascertain how these three factors increase Customer’s Intentions to Buy for sustainability in The Restaurant Industry. Structural Equation Modeling-Partial Least Squares (SEM-PLS), A Quantitative Method, is used in this study to examine the connection between variables. Data was gathered by randomly selecting 200 Customer’s Fast-Food Restaurants in The Jakarta Region. The Results indicate that while Online Reviews do not significantly affect Customer’s Intentions to Buy, Brand Images and Menu Visuals do. This implies that when it comes to Fast-Food Restaurants, Customers are more swayed by The Reputation of A Well-Known of Brand Images and Menu Visuals than by Online Reviews. For Restaurant Businesses looking to improve their Brand Images and optimize Their Menu Visuals to draw in and keep Customers, This Research offers Insightful Information.
Authors - Jawad Mahmood, Muhammad Adil Raja, John Loane, Fergal McCaffery Abstract - This research focuses on developing a testbed for Artificial Intelligence (AI) enabled Unmanned Aerial Vehicles (UAVs), particularly addressing the challenges posed by delayed rewards in Reinforcement Learning (RL). The Multi-UAV testbed integrates a realistic flight simulator with a Flight Dynamics Model (FDM), creating a versatile environment for testing and training RL algorithms. Two primary models were implemented: the Advantage Actor-Critic (A2C) model controls the target UAV, while the Asynchronous Advantage Actor-Critic (A3C) model governs the tracking UAV, leveraging asynchronous updates for efficient exploration and faster learning. A significant obstacle in reinforcement learning is the issue of delayed rewards, where feedback for an agent’s actions is not immediately available, potentially leading to unstable learning and reduced performance. This project addresses this challenge by integrating the Intrinsic Curiosity Module (ICM) with the A3C model. The ICM generates intrinsic rewards encouraging the A3C-controlled tracking UAV to explore new states, even in the absence of external rewards. This approach mitigates the effects of delayed rewards, allowing the tracking UAV to maintain effective pursuit of the target UAV under dynamic conditions.
Authors - Ufuk Sanver, Mustafa Cem Kasapbasi Abstract - Psoriasis, a chronic autoimmune skin condition, is an abnormal proliferation of skin cells along with scaling and inflammation. Proper detection and identification of psoriasis early on are critical to avoid inappropriate treatment and management. Traditional methods of diagnosis heavily rely on prolonged and subjective to physician experience of clinical presentation and histopathological study. Deep learning algorithms have proved highly successful for medical image analysis in the past few years and offer computerized, accurate, and quick diagnosis features. Here the use of convolutional neural networks (CNNs) is explored for classifying and detecting psoriasis from dermatology images. In the current research a pre-trained deep learning models that are ResNet50, MobileNetV2 and EfficientNet-B0 with SGDM, and DenseNet201 with ADAM optimization techniques have been employed. In this paper a curated dataset of psoriasis skin images affected with psoriasis was curated, in order to discriminate psoriatic lesions. To increase the accuracy dataset is augmented with various im-age processing techniques. In the training The experimental results demonstrate that the developed model is very accurate, sensitive, and specific. Our work suggests that deep learning models might be employed as valuable diagnostic aids, reducing errors in diagnosis and improved patient outcomes. Future studies will focus on enhancing model explain ability and increasing datasets with varied skin types and disease severities.
Authors - Diego Perez-Lopez, Rodolfo Bojorque, Jorge Duenas-Lerin, Fernando Ortega Abstract - Recommender systems play a crucial role in personalized content delivery, with collaborative filtering (CF) being a widely used approach. However, traditional CF methods often struggle to fully capture complex user-item interactions. In this study, we propose neural stacking models that integrate multiple CF techniques to enhance predictive accuracy. Experimental results show that, among baseline matrix factorization (MF) models, Biased MF and BNMF achieve the best Mean Absolute Error (MAE), demonstrating their effectiveness in modeling user-item relationships. Nonetheless, the proposed neural stacking models outperform these approaches by dynamically weighting CF models based on contextual factors. Comparisons with deep learning-based CF models (GMF, MLP, and NeuMF) confirm that neural stacking provides a more personalized and adaptive recommendation strategy. Future research will focus on optimizing model architectures, incorporating additional contextual information, and evaluating scalability for large-scale applications.
Authors - Ujjwal MK, Sunil Parameswaran, V. Guna Chowdary, Varun Bharadwaj, Suresh Jamadagni Abstract - In today’s fast-paced world, mental wellness is as essential as physical health. EmoHealth is a new mobile application designed to help individuals harmonize their mind and body. By combining emotion recognition with health analytics, EmoHealth delivers a real-time, holistic snapshot of a person’s mental and physical health. Using Apple technologies like HealthKit and CoreML, EmoHealth continuously tracks your heart rate, breathing patterns, and vocal activity through your Apple Watch and iPhone. This advanced approach detects even subtle shifts in emotional and physical health, sending timely alerts that promote self-care and help prevent stress-related issues from escalating. The EmoHealth app empowers users to stay mindful of their mental health while providing essential information to assist with stress management and emotional well-being. It keeps users updated on physical and emotional health metrics, fostering greater self-awareness and offering a new path to well-being. With on-device data processing, EmoHealth ensures both convenience and peace of mind, bridging gaps in traditional healthcare and enhancing proactive mental health support. This innovative approach highlights the potential of technology to deliver early, personalized feedback that can make a meaningful difference in people’s lives.
Authors - Motsotua Confidence Hlatshwayo,Kayode Oyetade,Tranos Zuva Abstract - In today’s rapidly evolving digital landscape, the ethical adoption of Artificial Intelligence (AI) has become a critical concern for organizations seeking to innovate responsibly. This study explores the integration of AI within South African Small and Medium Enterprises (SMEs), emphasizing the ethical challenges and strategic considerations associated with adoption in resource-constrained environments. While AI offers transformative potential for enhancing innovation, efficiency, and competitiveness, its deployment raises significant ethical risks, including algorithmic bias, data privacy violations, and a lack of transparency and accountability. Grounded in the Technology-Organization-Environment (TOE) framework and Ethical Decision-Making (EDM) theory, this research adopts a literature review methodology to examine the drivers, barriers, and ethical dimensions of AI adoption in SMEs. The study reveals that ethical awareness, organizational readiness, and external regulatory support are crucial to enabling responsible innovation. It highlights that SMEs with stronger ethical intent and governance practices are better positioned to adopt AI in ways that promote trust, fairness, and long-term sustainability despite resource limitations. This study contributes a context-sensitive conceptual framework that integrates structural and ethical considerations, offering practical guidance for policymakers, business leaders, and technology developers. The findings underscore the importance of aligning digital transformation with ethical imperatives to ensure inclusive and equitable AI adoption in developing economies.
Authors - Elrasheed Ismail Mohommoud Zayid, Abdulmalik A. Humayed, Yagoub Abbker Adam Abstract - This study aims to test machine learning (ML) classification analysis across a rich network traffic dataset sample to figure out the patterns of Saudi cyberspace and alert the community for cybersecurity risks. The network topology used for generating a sample dataset was a kind of heterogeneous simultaneous photonic multiprocessor exchange bus architecture (SOME-Bus). First, the dirty and noisy dataset was cleaned using essential cleaning procedures. Dimensions of 22 characteristics and 1048575 datapoints were considered for the model/data evaluation procedures. Second, the top-ranked learning model candidates were nominated by using the LazyPredict technique. Third, the Saudi cyber domain's terra-Byte payload traffic is shaped and visualized using a potent supervised computing algorithm. Powerful performance measure criteria were employed to calculate the model’s final accuracy and error rates. The findings conclude that Saudi was in the ninth rank with respect to cybercrime source; and, decision tree (DT) was the highest-performing algorithm with respect to destination address.
Authors - Vindiya Wickramathunga, Sapna Kumarapathirage Abstract - Accurate cinnamon plantation disease detection is crucial for planters, especially those with limited expertise. Traditional manual inspection methods are time-consuming and subjective, risking delays or incorrect diagnoses. This paper proposes an automated detection system using a Convolutional Neural Network (CNN) integrated with Explainable Artificial Intelligence (XAI) techniques. A publicly available dataset comprising three disease classes and one healthy class was used for training. The CNN achieved 75–80% accuracy with a 0.78 F1-score. Visual explanations via HiRes-CAM provided higher resolution compared to GradCAM++, enhancing user trust. The system demonstrated fast inference times (1–2 seconds), supporting real-time field deployment.
Authors - Leila BENOUNISSA, Djilali BENABOU, Soumia TABETI Abstract - The mobile phone is the most widely used mass communication object in the world. It gives its user the feeling of being everywhere at the same time and at all times (Jauréguiberry, 2005). Calls, ringtones and button clicks can be a source of undesirable phenomena for others and disrupt social interactions in public spaces (Palen, Salzman and Youngs, 2001). This is how several studies have attempted to understand the attitudes of mobile phone users and their preferences in different social settings. The objective of this contribution is to understand how users perceive the impact of the mobile phone use in different public spaces and if there are differences in the use compared to the characteristics of the users. Keywords: Mobile Phone, Social Settings, Attitudes, Algeria.
Authors - Shreya Joshi, Vijay Ukani, Priyank Thakkar, Mrudangi Thakker, Dhruvang Thakker Abstract - This study introduces a hybrid malware detection approach that combines machine learning (ML) and deep learning (DL) techniques to enhance detection accuracy. By applying models like K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Gradient Boosting, we achieved 99.47% accuracy during training and 99.21% accuracy during testing, focusing on analyzing Portable Executable (PE) header data. Additionally, incorporating Convolutional Neural Networks (CNN) with Long Short-Term Memory (LSTM) networks improved performance, achieving 99% accuracy, 97% precision, and 98% recall after 30 epochs. The proposed hybrid method reduces false positives and negatives while demonstrating scalability across various datasets, offering a reliable and efficient solution for contemporary malware detection.
Authors - Balvinder Pal Singh, Thangaraju B Abstract - Modern computing systems face the dual challenge of meeting escalating performance demands, driven especially by AI and ML workloads, while operating under stringent thermal constraints. As energy consumption continues to rise, conventional thermal mitigation techniques like frequency throttling often compromise performance and increase cooling costs, threatening long-term sustainability. This paper proposes a multi-level, software-centric approach to address this challenge through intelligent, temperature-aware scheduling. Leveraging evolutionary techniques, specifically Genetic Algorithm (GA), the proposed model reorders jobs at the OS scheduler level based on thermal impact and energy profiles. A secondary optimization phase further fine-tunes job execution using dynamic slice adjustment for thermally intensive tasks. Simulation results obtained using custom-integrated simulation framework leveraging GEM5, McPAT, and Hotspot tools, demonstrate 28% overall performance improvement, 53% reduction in thermal violations and a 15% decrease in energy consumption with 80% of the tasks executed without performance degradation. This approach was validated using representative benchmark workloads, optimizing both energy and temperature profiles. This AI-augmented, multi-level scheduling strategy significantly enhances thermal efficiency and performance, offering a scalable solution for next-generation high-performance computing environments.
Authors - Kostiantyn Hrishchenko, Oleksii Pysarchuk, Danylo Baran Abstract - This paper introduces a batch processing method for constraint programming to improve solution performance. The method is demonstrated in an example of a production scheduling problem. Transformations from mass production of standardized products toward small-scale, customized orders in the manufacturing sector introduce a new challenge of handling extensive input data. Modern production scheduling systems struggle to handle BigData loads caused by the limitations of state-of-the-art scheduling algorithms. Therefore, the development of highly adaptive algorithms and models capable of efficiently managing numerous unique orders while maintaining the ability to adapt to dynamic constraints and objectives becomes critically important. The baseline discrete constraint programming approach is chosen for its flexibility and extensibility, allowing it to model various realistic manufacturing scenarios. The proposed method splits large input into smaller subsets, each scheduled independently by repeatedly involving the constraint solver in different portions of the input data, significantly improving performance compared to allocating the whole input in a single step. Computational experiments with Google's OR-Tools CP-SAT solver evaluated the method's effectiveness. Time and memory usage reductions were shown. The proposed method demonstrates the possibility of solving problems of much larger size using the same constraint model and solver. It combines the advantages of the greedy algorithm and the exact integer programming approach.
Authors - V. Machaca, J. Grados, K. Lazarte, R. Escobedo, C. Lopez Abstract - The interaction between peptides and the Major Histocompatibility Complex (MHC) is a critical factor in the immune response against various threats. In this work, we fine-tuned protein language models like TAPE, ProtBert-BFD, ESM2(t6), ESM2(t12), ESM2(t30), and ESM2(t33) by adding a BiLSTM block in cascade for the task of peptide-MHC class-I binding prediction. Additionally, we addressed the vanishing gradient problem by employing LoRA, distillation, hyperparameter guidelines, and a layer freezing methodology. After experimentation, we found that TAPE and a distilled version of ESM2(t33) achieved the best results outperforming state-of-the-art tools such as NetMHCpan4.1, MHC urry2.0, Anthem, ACME, and MixMHCpred2.2 in terms of AUC, accuracy, recall, F1 score, and MCC.
Authors - Nafiz Eashrak, Mohammad Ikbal Hossain, Md Omum Siddique Auyon, Md Abdullah Al Adnan Abstract - The proliferation of cryptocurrencies and blockchain technology has significantly reshaped the financial sector, introducing decentralized and transparent digital transactions. Despite substantial advantages such as reduced transaction costs, enhanced transparency, and financial inclusion, the anonymous and decentralized nature of cryptocurrencies has also facilitated illicit financial activities, including fraud, money laundering, and tax evasion. This literature review systematically examines forensic methodologies, regulatory challenges, and theoretical frameworks relevant to cryptocurrency investigations. This study highlights advancements in blockchain analytics, AI-driven monitoring tools, and regulatory frameworks such as the FATF Travel Rule and EU MiCA. However, it also underscores persistent challenges posed by privacy-focused technologies, decentralized finance (DeFi), cross-border jurisdictional inconsistencies, and technical limitations in forensic methodologies. This paper proposes an integrated forensic framework incorporating AI analytics, international regulatory collaboration, specialized forensic training, and privacy-preserving investigative techniques. Through this comprehensive review, it provides critical insights for enhancing forensic investigation capabilities, regulatory compliance, and policymaking, while outlining future research opportunities addressing emerging threats in cryptocurrency-based financial systems.
Authors - Victor Sineglazov, Alexander Ruchkin Abstract - The article proposes a new combined approach to investigation the problem of classification on real-world noisy dataset using a multi-stage semi-supervised learning method. The main idea of this approach is based on combining two methods: self-supervised learning on unlabeled data using Contrastive Loss Nets and semi-supervised learning label propagation using an enhanced Poisson Seidel learning technique. The proposed approach offers significant advantages, as it allows for preliminary classification without labels, strengthening the distinctions between classes, and then using a minimal amount of labeled data for final classification. This is demonstrated through the analysis of synthetic data from different intersecting ”Two moons” and real medical dataset on heart disease - ”Cardio Vascular” Accuracy in the first case exceeds 82%, and for the second example - 73%, which is one of the best result on the Kaggle database when compared to any other known 20 methods.
Authors - Amir Aieb, Alexander Jacob, Antonio Liotta, Muhammad Azfar Yaqub Abstract - Predicting soil moisture under dynamic climate conditions is challenging due to intricate dependencies within the data. This study presents a surrogate deep learning (SDL) model with a multitask learning (MTL) approach to improve daily soil moisture predictions across spatiotemporal scales. The model employs a two-level encoding process, first compressing climate parameters into a single feature and then applying sequential encoding to capture long-term temporal patterns within a one-year timeframe for better generalization. Seasonality detection using autocorrelation facilitates data resampling into homogeneous samples, enhancing the SDL model by optimizing hyperparameters through efficient weight sharing between layers. To evaluate the effectiveness of MTL, three different SDL architectures such as LSTM, ConvLSTM, and BiLSTM were implemented for a comprehensive analysis. All models struggle with soil moisture prediction, particularly during dry periods, where LSTM experiences the most significant accuracy drop. While BiLSTM demonstrates better performance, its effectiveness remains constrained. However, integrating MTL enhances model stability and spatio-temporal accuracy, reducing errors across various conditions and achieving a 10% improvement due to better data representation, enabling SDL models to capture regional heterogeneity more effectively.
Authors - V. Machaca, D. Lopez, J. Mamani, S. Ramos, Y. Tupac Abstract - Cancer immunotherapy offers a promising option to conventional cancer therapies, in this field, neoantigen detection is a rapidly evolving field; however, as it requires multiple bioinformatics stages, such as quality control, alignment, variant calling, annotation, and neoantigen prioritization. Each stage depends on specific software tools, which can create technical and compatibility challenges, often requiring significant expertise to integrate and manage. To address these challenges, we introduce NeoArgosTools, a novel flowchart-based platform, with an intuitive graphical interface, designed to simplify and streamline neoantigen detection pipelines in cancer immunotherapy research.
Authors - Fhatuwani Mapande, Tranos Zuva, Kayode Oyetade Abstract - This paper presents a comprehensive and integrative review of key user perception models in the context of technology adoption. It critically examines five influential frameworks like Technology Acceptance Model (TAM), Unified Theory of Acceptance and Use of Technology (UTAUT), Diffusion of Innovation (DOI), Innovation Adoption Lifecycle Model, and Technology Readiness Model (TRM) to uncover the multidimensional factors shaping user attitudes and behaviors toward emerging technologies. Through a structured comparative analysis, the study explores core constructs such as perceived usefulness, ease of use, social influence, user readiness, and psychological traits including optimism and discomfort. The novelty of this work lies in its synthesis of diverse theoretical perspectives, offering a holistic view that bridges cognitive, emotional, and sociocultural dimensions of technology adoption. The findings underscore the significance of integrated, user-centered approaches and highlight the role of contextual and sector-specific variables in influencing adoption outcomes. Practical recommendations are provided for researchers, developers, educators, and policymakers to design inclusive and adaptive strategies that enhance technology acceptance and sustained engagement. This study contributes to advancing theoretical understanding and guiding practical interventions in the evolving landscape of digital transformation.
Authors - Staphord Bengesi, Hoda El-Sayed, Md Kamruzzaman Sarker Abstract - Alzheimer's and Parkinson's diseases are two progressive neurodegenerative disorders that primarily affect senior citizens worldwide, and currently, there is no cure. In recent years, the number of diagnosed cases has been increasing. Since both diseases have an impact on the brain, MRI images are used as a crucial diagnostic tool. With advancements in AI, machine learning models are showing great promise in diagnosing and classifying MRI images. To explore this potential, we developed and tested five transformer models, such as ViT, Swin, DeiT, MedT, and Swin-ViT, using a Kaggle dataset containing MRI images from individuals with Alzheimer’s, Parkinson’s, and healthy controls. The models were evaluated on both a balanced dataset of over 2,900 samples and an unbalanced dataset of more than 7,000 samples. Our findings revealed that models trained on the unbalanced dataset outperformed those trained on the balanced dataset, highlighting the advantage of larger datasets in enhancing model performance.
Authors - Zuzana Trabalkova, Martin Stevik, Kamil Zelenak, Jakub Dandar, Zdenek Straka, Daniel Kvak, Karolina Kvakova, Petra Ovesna Abstract - The growing demand for chest radiography in healthcare, combined with radiologist shortages and increasing workloads, underscores the need for innovative diagnostic support tools. This crossover study evaluates the effect of commercially available deep learning-based automatic detection software (DLAD) on radiologists’ diagnostic performance in chest X-ray (CXR) interpretation. Five radiologists independently assessed a dataset of 540 anonymized CXRs, both independently and with DLAD assistance, in two phases separated by a 30-day washout period. DLAD assistance significantly improved diagnostic performance, with overall sensitivity (Se) increased from 0.762 (95% CI: 0.705–0.811) to 0.911 (0.870–0.941, p < 0.001), while specificity (Sp) remained unchanged at 0.850 (0.805–0.887, p = 0.331). The positive predictive value (PPV ) slightly improved from 0.810 (0.755–0.856) to 0.836 (0.788–0.876, p = 0.331), and the negative predictive value (NPV ) increased from 0.810 (0.763–0.850) to 0.941 (0.882–0.947, p < 0.001). These improvements were consistent across radiologists, with notable reductions in false-negative rates. The findings emphasize DLAD’s potential to standardize diagnostic accuracy, enhance sensitivity, and support radiologists in chest X-ray interpretation. These results highlight the clinical value of AI-assisted workflows in improving detection rates while maintaining specificity.
Authors - Prince Kelvin Owusu, Caleb Annan, Ruhiya Abubakar, Moses Aggor, Emelia Sarpong, Gibson Afriyie Owusu, Jefferson Oduro Asiamah, Martins Larweh Nuertey Abstract - This study explores the market feasibility, challenges, and prospects of integrating mini-grids in island communities on Ghana’s Volta Lake, utilizing an exploratory sequential mixed-method design. Purposive and census sampling techniques were employed for qualitative and quantitative research, respectively, with 65 participants. Thematic analysis was applied to qualitative data from semi-structured interviews, while the Analytic Hierarchy Process (AHP) assessed Likert scale-based questionnaire responses. Identifying 21 challenges categorized into economic, political, technical, environmental, and social classes, economic challenges ranked highest (38.64%), with access to nance as the most significant challenge (12.03%). Despite a viable market, the study highlights a potential decrease in donor funding for mini-grid development in Ghana. Significantly, it concludes that policy unsuitability has cascading effects, necessitating a modification in the approach to enhance minigrid development in Ghana, emphasizing policy, tari scheme, and business model adjustments for holistic improvement.
Authors - Prince Kelvin Owusu, Philimina Pomaah Ofori, Moses Aggor, Gibson Afriyie Owusu, Jefferson Oduro Asiamah, Martins Larweh Nuertey, Joseph Djossou Akwetey, Joel Nana Sarfo Konadu Abstract - The integration of immersive technologies such as Augmented Reality (AR) and Virtual Reality (VR) is revolutionizing global marketing and advertising strategies, yet their application within Ghana’s communication landscape remains underexplored. This study investigates the pioneering role of AR and VR in transforming marketing practices in Ghana, with a focus on how these technologies influence consumer engagement, brand perception, and strategic communication. Employing a mixed-methods research design, the study combines qualitative interviews with 15 marketing professionals across major urban centers and quantitative survey data from 250 consumers who have interacted with immersive advertisements in retail, real estate, and tourism sectors. The findings reveal a rising trend in experimental AR/VR adoption among Ghanaian firms, driven by a desire to differentiate brands and deepen customer interaction. However, results also indicate significant barriers, including high implementation costs, limited technological infrastructure, and a lack of skilled personnel. Consumer responses demonstrated high engagement and positive emotional reactions to AR/VR content, particularly among younger demographics, though accessibility concerns persist. The study concludes that while AR and VR offer transformative potential for Ghana’s marketing sector, their long-term success depends on strategic investment in digital infrastructure, public-private partnerships, and targeted capacity-building programs. It recommends that policymakers support immersive technology adoption through subsidies and training initiatives, while marketers should focus on culturally relevant, mobile-optimized AR/VR campaigns to maximize reach and effectiveness. This research contributes to the growing discourse on digital innovation in emerging economies and provides a roadmap for integrating immersive technologies into Ghana’s evolving communication ecosystem.
Authors - Fawzy Alsharif, Irem Yildirim Abstract - Microwave imaging is a promising non-invasive technique for early stage cancer detection, leveraging its sensitivity to variations in the dielectric properties of biological tissues. In this work, a compact ultra-wideband (UWB) antenna specifically designed for lung cancer imaging is presented and analyzed through electromagnetic simulations. The antenna is fabricated on a Rogers RT5880 substrate (εr = 2.2, thickness = 1.65 mm) with overall dimensions of 23 × 21 × 1.58 mm³ and is impedance-matched to a 50 Ω feedline. Performance evaluations using CST and HFSS reveal operation across three frequency bands centered at 3.08 GHz, 6.04 GHz, and 9.54 GHz. The antenna achieves a peak gain of 4.52 dBi and a maximum radiation efficiency of 86% at the highest frequency. It offers a wide operational bandwidth from 2.58 GHz to 11.67 GHz. A realistic lung phantom modeled in CST demonstrates the antenna’s effectiveness in detecting signal changes caused by dielectric contrast in tissues, highlighting its potential for accurate and non-invasive lung cancer diagnosis.
Authors - Manideep Pendyala, Udit Goel, Jim Samuel, Pal Patel, Janki Kanakia, Alexander Pelaez, Neel Savalia, Tanya Khanna Abstract - Emojis have become an integral part of modern digital communication. Despite their widespread use, most sentiment analysis methods and models disregard emojis during preprocessing, leading to the loss of vital emotional cues. This paper introduces a curated dataset of sentence pairs, with and without emojis, each annotated across three sentiment categories, to assess the impact of emoji inclusion on sentiment classification. We evaluate emoji-inclusive and emoji-exclusive strategies against our human-determined gold standard, using a range of approaches, including the traditional lexicon-dictionary based methods, and also artificial intelligence (AI) methods including pre-trained machine learning (ML) based classifiers, and large language models (LLMs). Results show that retaining emojis significantly enhances the performance of all the LLMs we tested, with models such as Qwen, Deepseek, Bert and Mistral achieving accuracy improvements of over 25%, over an emoji-exclusive strategy. These findings highlight that emojis carry meaningful semantic and affective signals. We emphasize the limitations of current approaches to emoji handling, where emojis are often ignored or treated as irrelevant noise. Instead, we advocate for more thoughtful methods that recognize emojis as meaningful components of communication and incorporate them as valuable sources of information.
Authors - Diego Ricardo Salazar-Armijos, Hector Mauricio Revelo-Herrera, Holger Alfredo Zapata-Mayorga, Paul Diaz-Zuniga, Aida Noemy Bedon-Bedon, Nelson Fernando Vinueza-Escobar Abstract - This research, conducted within the framework of the project "Dropout in Higher Education – Early Warning Model with Emerging Technologies at the University of the Armed Forces ESPE", analyzed the factors influencing the dropout of Information Technology students at the Santo Domingo campus between 2017 and 2023. Socioeconomic and academic variables were considered, based on enrollment data in accordance with the regulations of higher education in Ecuador. Logistic regression and decision tree algorithms were applied due to their classification capabilities and statistical relevance. Additionally, an ANOVA-based comparison was performed. The study concluded that academic performance is the main factor associated with student dropout.
