International Journal of Artificial Intelligence & Mathematical Sciences

Decomposition of CI and ENSO signals using the wavelet-Based technique

2026-08-07T07:35:05+00:00

The study is focused on six solar cycles that are studied in a comparison of related ENSO signals by using the wavelet technique. The wavelet technique is a advanced method used to analyze and understand the dynamical behavior of signals in any dimension. In this work, Discrete wavelet transformation (DWT) is applied on sun and ENSO data signals. Using the Daubechies wavelet (db 2), the Solar and ENSO parameters are decomposed into their approximation (a) and detail (d) components. By this technique, dynamical changes in both studied parameters are found effectively. The variation of solar energy was observed at a different level detailed coefficient frequencies up to 5 levels. The decomposition at level 5 shows the original signal and lower variability. Further initial and middle phase solar variation was found in the change of ENSO variability. The variability of solar energy in contrast to climate signals will be helpful in the future to understand the dynamical signal process by wavelet transformation.

Generalized Inequalities of Hermite-Hadamard Like for (m_1,m_2)−Convex Function Via Ho ̈lder & Power-Mean Inequalities and Implementations

2026-08-07T07:35:05+00:00

In the present article, we prove the generalized Hermite-Hadamard (H-H) like inequalities for −convex (conv.) via H lder & Power-mean inequalities and implementations for theory of probability (prob.) & numerical integration are deduced. Some consequences of several published articles would be captured as especial cases. Moreover, we deduce few especial cases of −conv. function (func.).

Transforming Tuberculosis Care: The Role of Deep Learning Models in MDR-TB Prediction

2026-08-07T07:35:05+00:00

Multidrug-resistant TB(MDR-TB), a subtype of TB that remains a global health issue, is a significant barrier to detection and treatment. With a focus on MDR-TB prediction and early identification, this paper investigates the revolutionary implications of AI and ML in TB treatment. This research uses a range of sources to explore possible risks, demographic integration, and the performance of ML algorithms. The literature review focuses on the prospects of DL for TB-diagnosis, covering current breakthroughs, challenges, and resources like UNet and TBXNet. Critical reviews examine the generalizability of AI models, ethical implications, and cost-effectiveness. The study highlighted many important tasks, including exploring demographic integration, examining financial ramifications, and developing optimal ML models. Considering ethical issues and overcoming difficulties in various healthcare settings, the findings suggest that AI and ML may be effective in TB detection.

Cloud-Enabled IoT Framework for Induction Motor Fault Monitoring

2026-08-07T07:35:05+00:00

Induction motors are employed in several applications, such as the powering of large motor equipment in industry, as well as smaller ones such as household appliances. They operate pumps, fans, compressors in industries and are used in elevators, heating, ventilation, and air conditioning (HVAC) systems, and even in some electric vehicles. This study introduces a solution to monitor the health status of induction motor by implementing an IoT based fault detection and continuous monitoring system, which are considered critical assets in various industrial and commercial applications. Traditional maintenance approaches are linked to un-necessary premature expenses, low productivity and equipment failures. This system helps overcome these shortcomings by taking advantage of the Internet of Things power to implement predictive maintenance policies. The proposed system involves a web of strategically located sensors (vibration, temperature, current, and voltage) to constantly obtain important data on the operation of the equipment (fault detection and constant health control). This system presents an IoT-based system that can be used to monitor the induction motors, which are important in many industrial and commercial processes, for faults and health monitoring. These data are transmitted to the IoT gateway of a cloud-based system where 1) advanced data processing and analysis is used to process the data and analyze it and 2) machine learning algorithms are used to detect anomalies and identify the pattern of faults. The system is designed to detect many common motor issues, including imbalance, winding and bearing degradation, early. One of the outputs will be a real-time dashboard for the motor health status, performance trends and actionable insights. Moreover, the system produces automatic notifications and alerts to the assigned staff when it detects unusual motor activity or an imminent failure, which allows a timely response. This IoT solution will significantly reduce unscheduled downtime, maintenance costs and motor life, improve operational efficiency and plant safety, and contribute to the development of an intelligent and resilient industrial environment. In the Future Maximum will be achieved by combining Edge AI and Digital Twins for real-time analysis at low latency on-site. This allows fault prognosis and hyper-efficient and predictive maintenance scheduling to occur in real time, reducing downtime and costs.

Lightweight Post Quantum Homomorphic Encryption for Secure Network Traffic Intrusion detection

2026-08-07T07:35:05+00:00

These quantum computing technologies are a serious risk for existing “Cryptographic Algorithms, corresponding “RSA, ECC, and Diffie-Hellman”. The Quantum Algorithm “Shor's can break classical” public-key encryption systems effectively and now there is a momentous security concern of cloud computing, IoT systems, and healthcare networks, as well as the future 6G communication systems. In response to the above difficulties, researchers have suggested two new methods of privacy-preserving encrypted computation, namely: “Post-Quantum Cryptography (PQC) and Fully Homomorphic Encryption (FHE)". But the current “Post-Quantum Homomorphic Encryption (PQHE)” solutions have high computational complexity, expanded ciphertext size, latency, and are not widely deployed in cybersecurity applications. This research aims to provide a lightweight Post-Quantum Homomorphic Encryption framework for secure network traffic analysis with the intrusion detection dataset from CICIDs 2018. The proposed framework combines the lattice-based PQHE concepts, “Principal Component Analysis (PCA) and K-Means cluster to envision encrypted traffic and secure intrusion analysis. Data preprocessing, feature standardization, dimensionality reduction, encrypted traffic representation, clustering analysis and graphical visualization are parts of the experimental methodology. To address the key challenges, by using two major approaches for privacy-preserving encrypted computation have been proposed: “Post-Quantum Cryptography (PQC) and Fully Homomorphic Encryption (FHE)”. The current Post-Quantum Homomorphic Encryption (PQHE) systems, on the contrary, are complicated, costly in the length of the ciphertext, slow and not widely used in terms of cybersecurity. To protect network traffic analysis a lightweight Post-Quantum Homomorphic Encryption framework is proposed in this research to achieve network security using intrusion detection data set CICIDS2018.

Real-Time Face Mask Detection Using Transfer Learning with MobileNetV2

2026-08-07T07:35:05+00:00

The COVID-19 pandemic created an urgent need for automated systems capable of verifying face-mask compliance in public spaces. This paper presents a lightweight, real-time face-mask classifier built on transfer learning with the MobileNetV2 architecture. A pretrained ImageNet backbone is used as a frozen feature extractor with a compact classification head, followed by a fine-tuning phase that unfreezes the final convolutional layers. The model is trained and evaluated on the publicly available. Face Mask ∼12K Images dataset, comprising approximately twelve thousand pre-cropped and class-balanced face images split into training, validation, and test partitions. Using data augmentation, two-phase training, and standard regularization, the classifier attains approximately 99% accuracy on the held-out test set with near-perfect precision and recall for both the masked and unmasked classes. The results confirm that a low-compute, mobile-oriented backbone combined with transfer learning is sufficient for accurate binary mask detection, making the approach suitable for deployment on edge devices. The proposed pipeline is a clean, reproducible, end-to-end implementation rather than a novel methodology.