Automated medical diagnosis is an important topic, especially in detection and classification of diseases. Malaria is one of the most widespread diseases, with more than 200 million cases, according to the 2016 WHO report. Malaria is usually diagnosed using thin and thick blood smears under a microscope. However, proper diagnosis is difficult, especially in poor countries where the disease is most widespread. Therefore, automatic diagnostics helps in identifying the disease through images of red blood cells, with the use of machine learning techniques and digital image processing. This paper presents an accurate model using a Deep Convolutional Neural Network build from scratch. The paper also proposed three CNN models each one trained on the Malaria RBC dataset with different architectures for handling the classification tasks. Furthermore, disadvantage of the traditional method of using transfer learning, and how to control model complexity to achieve better performance was discussed. The dropout regularization technique was used to avoid overfitting problems and minimize validation loss. Applying Data Augmentation technique to avoid the problem of small data in training of proposed models, which is a very common problem in medical dataset. Finally, removing noise in Malaria images using a Median blur filter, and studying how effects of that on training CNN models. According to the classification results, the proposed model achieved better classification results at accuracy 99.22 on the original Malaria RBCs dataset, and it has the best performance comparing with related work.

Malaria is a contagious disease that infects many people each year. It is diagnosed in the laboratory using an expert and thorough examination of healthy and si Malaria is a contagious disease that infects many people each year. It is diagnosed in the laboratory using an expert and thorough examination of healthy and si

Federated Learning FL is a rapidly evolving machine learning paradigm that enables collaborative model training across decentralized data sources while preserving data privacy. Since its inception in 2016, FL has emerged as a transformative approach in domains such as healthcare, IoT, and edge computing, where data sensitivity and regulatory constraints limit centralized processing. This systematic review consolidates findings from 50 high-quality studies selected from over 250 papers to present a comprehensive synthesis of FL methodologies, core aggregation techniques, privacy-preserving mechanisms, security threats, domain-specific applications, and emerging trends. We categorize challenges into communication overhead, heterogeneity (device, data, model), fairness, trust, and evaluation inconsistencies. We highlight research gaps, notably in standardized evaluation, incentive mechanisms, and deployment scalability. By reviewing recent advances such as vertical federated learning, federated unlearning, and blockchain-based incentives, this paper offers a roadmap for future research and identifies open questions vital for widespread FL adoption.

In disaster scenarios where traditional communication infrastructure is unavailable or damaged, the rapid deployment of efficient ad hoc wireless networks is critical. These networks must be adaptable, self-organizing, and capable of maintaining reliable communication links under uncertain and dynamic environmental conditions. However, obtaining optimal ad hoc network design and accurate node localization remains a challenging task, especially in disaster scenarios where time is critical. This work addresses these issues by introducing a novel Multi-Conditioning Variational Autoencoder (MCVAE) model for generating optimized ad hoc node localization conditioned on multiple factors, such as network area size, node count, and success criteria. The proposed generative model not only addresses the localization problem but also enables on-demand generation of emergency ad hoc networks, making it a practical solution for real-time deployments in disaster-stricken areas. Furthermore, a synthetic dataset incorporating realistic features that effectively describe the ad hoc node distribution environment was created. The MCVAE model was trained on this dataset to learn complex dependencies between input conditions and network design. Experimental results demonstrate that the model effectively generalizes beyond the training area sizes, accurately generating node layouts for previously unseen dimensions with efficient distribution and low localization error (RMSE 0.125 for a network size of 450 m2 with 285 nodes) and an R2 score of 0.9999. Additionally, the proposed methodology achieves excellent scalability and adaptability for generative ad hoc networks, allowing users to specify different area dimensions and automatically receive efficient node distributions with minimal localization error. The findings highlight the model's potential for real-time, data-driven deployment of resilient ad hoc networks in disaster-stricken or infrastructure-limited regions. This is the first work to contribute a practical and scalable use of a generative AI model that can significantly enhance wireless communication readiness in emergency situations.

Detecting the validity of fruits and vegetables is one of the major challenges tasks in industrial agriculture applications, especially in the canning and fruit picking and isolating processes that are utilized for exporting. It involves subjective assessment by the human judgment of the ripeness stage of fruits for fresh produce, which is often time-consuming and inconsistent. As automation continues in all industrial and agricultural fields, there is an increasing need for a reliable method to handle this process. This paper proposes an automated system using convolutional neural networks to classify green apples concerning their ripeness level. Our approach lies in introducing a new lightweight modification of the VGG-16 architecture. Traditional VGG-16 models, while effective, are computationally intensive and may not be suitable for deployment in resource-constrained environments such as embedded systems. The proposed Lightweight-VGG significantly reduces the model size, and trainable parameters while achieving the best classification accuracy. The proposed model was tested using 1000 samples of apple fruit images and achieved an accuracy of 98.25%, compared to 96.8% achieved by the original VGG-16. The proposed model also shows improvements in precision, recall, and F1-score, making it a viable alternative for real-time fruit quality assessment applications and constrained devices.