Pneumonia remains a significant global health concern, necessitating efficient diagnostic tools. This study presents a novel Convolutional Neural Network (CNN) architecture, CuDenseNet, designed for the binary classification of Chest X-Ray (CXR) images as either having pneumonia or normal (healthy). Unlike models that rely on transfer learning from pre-trained architectures, CuDenseNet is trained from scratch and incorporates three parallel DenseNet paths of varying depths, enhancing feature extraction and classification accuracy. The model was evaluated on a combined dataset of 11,708 CXR images, achieving exceptional performance metrics of 99.1% accuracy, 99.7% precision, 99.1% recall, and an AUC of 99.7%. The comparative analysis demonstrates that CuDenseNet outperforms state-of-the-art pre-trained models such as VGG19 and ResNet50 while providing superior adaptability. These results underscore the potential of CuDenseNet as a robust and reliable tool for automated pneumonia diagnosis, with significant implications for clinical applications and future research in medical imaging.

This study addresses the challenge of effectively modeling uncertainty and hesitation in complex decision-making environments, where traditional fuzzy and vague set models often fall short. To overcome these limitations, we propose the Fermatean neutrosophic vague soft set (FNVSS), an advanced extension that integrates the concepts of neutrosophic sets with Fermatean membership functions into the framework of vague sets. The FNVSS model enhances the representation of truth, indeterminacy, and falsity degrees, providing greater flexibility and resilience in capturing ambiguous and imprecise information. We systematically develop new operations for the FNVSS, including union, intersection, complementation, the Fermatean neutrosophic vague normalized weighted average (FNVNWA) operator, the generalized Fermatean neutrosophic vague normalized weighted average (GFNVNWA) operator, and an adapted Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) method. To demonstrate the practicality of the proposed methodology, we apply it to a solar panel selection problem, where managing uncertainty is crucial. Comparative results indicate that the FNVSS significantly outperforms traditional fuzzy and vague set approaches, leading to more reliable and accurate decision outcomes. This work contributes to the advancement of predictive decision-making systems, particularly in fields requiring high precision, adaptability, and robust uncertainty modeling.

An adaptive e-learning environment faces significant challenges in offering personalized learning resources for Deaf and Hard-Hearing (DHH) learners. These learners exhibit diverse preferences in learning and communication, influenced by their characteristics related to deafness, highlighting the need for personalized educational content. A well-defined learning model is essential to map the characteristics of learners to suitable learning resources, enabling effective recommendations within an e-learning system. This study explores the development of a comprehensive DHH learner model, focusing on the presence of multiple learning preferences based on the VARK (Visual, Aural, Read/Write, and Kinesthetic) learning style model and the effectiveness of fuzzy clustering in capturing the diverse but overlapping preferences. Fuzzy-C-Means (FCM) successfully identified six different but overlapping clusters, indicating that most learners exhibit multimodal learning preferences rather than relying solely on a visual learning style. Cluster centroid analysis reveals that the visual learning style is the most preferred, while aural learning is the least favored among DHH learners. By calculating the overall learning style score based on the fuzzy membership value across all clusters on all four dimensions of VARK, learners' learning style preferences were validated against self-reported data. The evaluation involved a survey of 130 higher secondary DHH students from Kerala, India, yielding promising results (precision: 0.90, recall: 0.84, F1-score: 0.84) on the model's efficiency in identifying the dominant learning style. These findings emphasize the need for adaptive content delivery strategies that integrate text, visual, and interactive elements to enhance the engagement of DHH learners. However, the limited sample size, due to the unavailability of publicly accessible datasets, and the limited number of students in higher secondary education, further highlights the need for accessible and standardized DHH data to advance this research domain.

This paper proposes a new digital signature algorithm that employs the Heisenberg group over a prime field and exploits its structure to provide strong cryptographic properties. The developed scheme involves the Heisenberg group within a commutator-based authentication system contributing to key exchange while providing secure means of authentication as well as data integrity. The proposed method for digital signature generation and verification is described by a comprehensive algorithm, which proof of correctness is provided by a confirmation theorem. The security analysis establishes the strength of the scheme against different cryptanalytic attacks and proves the effectiveness of non-repudiation and message integrity in the digital environment. The efficiency of the presented security scheme is also evaluated in practice, using the finite field Z₁₁, where it is measured against other standards such as ECDSA and RSA. While it has a bigger signature size, the scheme is still efficient when it comes to signing and verifying messages which highlights its suitability for protecting communications under post-quantum cryptography. It supports the on-going progress of cryptography, giving added security to new technologies when faced with the threat of complex attacks.