Welcome to AfricArXiv
This initiative showcases UbuntuNet's commitment to fostering knowledge sharing, collaboration, and accessibility within the African research community. With AfricArxiv, researchers across the continent have a dedicated platform to disseminate their findings, making them accessible to a global audience. By facilitating open access to scholarly work, UbuntuNet Alliance plays a pivotal role in advancing the principles of open science, enhancing research visibility, and driving innovation across Africa.
Communities in AfricArxiv
Select a community to browse its collections.
- The general repository is open for individual submissions by researchers, librarians and research administrators.
- Showcase of project activities, presentations, and scholarly contributions curated by the AfricArXiv initiative.
- A Rapid Grant Fund to address research questions and implement science engagement activities associated with COVID-19
- An initiative to support the development of a harmonised quality assurance and accreditation system at institutional, national, regional and Pan-African continental level.
- Facilitating knowledge sharing and collaboration among institutions, researchers, and educators within the Ubuntunet Alliance network.
Recent Submissions
CERN–UNESCO–NFR School on Open Science
(NRF, 2025-02-17) NRF
HYBRID NEURAL NETWORK MODELS FOR PRIVACY-PRESERVING DATA PROCESSING
(2025)
In the era of big data, the need to balance efficient data processing with robust privacy safeguards has become increasingly critical. This paper explores Hybrid Neural Network Models as a promising solution for privacy-preserving data processing. By integrating the strengths of traditional neural architectures (such as Convolutional Neural Networks and Recurrent Neural Networks) with privacy-enhancing technologies like differential privacy, federated learning, and homomorphic encryption, hybrid models can achieve high performance without compromising sensitive information. We discuss architectural designs that optimize both data utility and privacy protection, focusing on adaptive learning mechanisms that minimize data exposure during training and inference. Experimental evaluations across diverse datasets demonstrate that these models maintain competitive accuracy while significantly reducing privacy risks. This research highlights the potential of hybrid neural networks to enable secure, scalable, and efficient data processing in privacy-sensitive domains such as healthcare, finance, and IoT.
APPLICATIONS OF CONVOLUTIONAL NEURAL NETWORKS (CNNS) IN MEDICAL IMAGE SECURITY
(2025)
The rapid growth of digital healthcare systems has led to an increasing reliance on medical imaging for diagnosis and treatment, raising significant concerns regarding data security and patient privacy. Convolutional Neural Networks (CNNs), a class of deep learning algorithms renowned for their exceptional performance in image analysis, have emerged as powerful tools for enhancing medical image security. This abstract explores the diverse applications of CNNs in securing medical images, including encryption, authentication, watermarking, and anomaly detection. CNNs can learn complex patterns and features, enabling robust image encryption techniques that protect data against unauthorized access and tampering. Additionally, CNN-based watermarking ensures data integrity and authentication without compromising image quality, while their ability to detect subtle anomalies helps identify potential security breaches. The integration of CNNs in medical image security not only improves data protection but also ensures compliance with stringent healthcare regulations, safeguarding patient confidentiality in the digital era. This paper highlights recent advancements, challenges, and future directions in leveraging CNNs for comprehensive medical image security solutions.
DEEP NEURAL NETWORKS FOR BIOMETRIC AUTHENTICATION IN INFORMATION SECURITY
(2025)
In the rapidly evolving landscape of information security, biometric authentication has emerged as a critical mechanism for enhancing identity verification processes. Deep Neural Networks (DNNs), with their exceptional capability to learn complex patterns and representations from large datasets, have significantly advanced the accuracy and robustness of biometric systems. This paper explores the application of DNNs in biometric authentication, focusing on modalities such as fingerprint recognition, facial recognition, iris scanning, and voice authentication. We discuss the architectural innovations in deep learning, including Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), that contribute to improved feature extraction, noise reduction, and resistance to spoofing attacks. Furthermore, the paper addresses the challenges associated with deploying DNN-based biometric systems, such as data privacy concerns, model interpretability, and susceptibility to adversarial attacks. By analyzing recent advancements and experimental results, we demonstrate how DNNs enhance the security, scalability, and efficiency of biometric authentication in modern information security infrastructures.