Epilepsy is a prevalent neurological disorder with considerable risks, including physical impairment and irreversible brain damage from seizures. Given these challenges, the urgency for prompt and accurate seizure detection cannot be overstated. Traditionally, experts have relied on manual EEG signal analyses for seizure detection, which is labor-intensive and prone to human error. Recognizing this limitation, the rise in deep learning methods has been heralded as a promising avenue, offering more refined diagnostic precision. On the other hand, the prevailing challenge in many models is their constrained emphasis on specific domains, potentially diminishing their robustness and precision in complex real-world environments. This paper presents a novel model that seamlessly integrates the salient features from the time–frequency domain along with pivotal statistical attributes derived from EEG signals. This fusion process involves the integration of essential statistics, including the mean, median, and variance, combined with the rich data from compressed time–frequency (CWT) images processed using autoencoders. This multidimensional feature set provides a robust foundation for subsequent analytic steps. A long short-term memory (LSTM) network, meticulously optimized for the renowned Bonn Epilepsy dataset, was used to enhance the capability of the proposed model. Preliminary evaluations underscore the prowess of the proposed model: a remarkable 100% accuracy in most of the binary classifications, exceeding 95% accuracy in three-class and four-class challenges, and a commendable rate, exceeding 93.5% for the five-class classification.

Millions of people lose their lives each year worldwide due to traffic law violations, specifically, over speeding. The existing systems fail to report most of such violations due to their respective flaws. For instance, speed guns work in isolation and cannot measure speed of all vehicles on roads at all spatial points. They can only detect the speed of the vehicle the line of sight of the camera. A solution is to deploy a huge number of speed guns at different locations on the road to detect and report vehicles that are over speeding. However, this solution is not feasible because it demands a large amount of equipment and computational resources to process such a big amount of data. In this paper, a speed detection framework is developed to detect vehicles’ speeds with only two speed guns, which can report speed even when the vehicle is not within the camera’s line of sight. The system is specifically designed for an irregular traffic scenario such as that of Pakistan, where it is inconvenient to install conventional systems. The idea is to calculate the average speed of vehicles traveling in a specific region, for instance, between two spatial points. A low-cost Raspberry Pi (RPi) module and an ordinary camera are deployed to detect the registration numbers on vehicle license plates. This hardware presents a more stable system since it is powered by a low consumption Raspberry Pi that can operate for hours without crashing or malfunctioning. More specifically, the entrance and exit locations and the time taken to get from one point to another are recorded. An automatic alert to traffic authorities is generated when a driver is over speeding. A detailed explanation of the hardware prototype and the algorithms is given, along with the setup configurations of the hardware prototype, the website, and the mobile device applications.

Rotary machine fault diagnosis is essential for maintaining system performance and ensuring operational safety. Deep data–driven models have demonstrated exceptional performance in fault diagnosis of rotary machines when sufficient training data is available. However, obtaining this data, particularly under faulty conditions, can be labor-intensive and time-consuming. The challenges posed by a data-driven model with insufficient training data have been addressed by transfer learning (TL) techniques when similar machine data is available. However, these conventional TL-based models struggle to perform effectively for a machine where gathering similar training data is difficult. This paper presents a deep TL approach with an optimized fine-tuning strategy, enabling the effective transfer of knowledge from a source domain with diverse machine data to a target domain with limited data from target machines. The proposed framework introduces a feature-space guided fine-tuning strategy in which the optimal retraining depth is determined through quantitative evaluation of target-domain feature separability across successive pretrained layers using silhouette score analysis. This transforms conventional heuristic layer-freezing into an objective data-driven adaptation process, enabling controlled transfer under domain shift and limited target data availability. Experimental results using data obtained from a real industrial testbeds involving milling cutting tool, grinding machine and public Paderborn dataset validate the effectiveness of the proposed approach and demonstrate its superiority over existing TL techniques in the similar domain.

In today’s digital age, coverage prediction is essential for optimizing wireless networks and improving user experience. While numerous path loss models and advanced machine learning algorithms have been developed to achieve high prediction performance, they predominantly operate within a centralized learning paradigm. While effective, this conventional approach often suffers from scalability and privacy limitations that are critical to the successful deployment of wireless maps. Conversely, in this paper, we propose a novel decentralized approach based on a federated learning long short-term memory (LSTM) model to accurately predict network coverage in indoor environments. The proposed FedLSTM is a method that allows multiple users, or clients, to train the model without sharing their personal data directly with a central server. In an experimental setup, we used real data collected from numerous clients moving along different paths. The FedLSTM model is evaluated in terms of root mean square error (RMSE), mean absolute error (MAE), and R2. Furthermore, compared to a centralized counterpart, FedLSTM shows a slight increase in RMSE from 2.4 dBm to 2.5 dBm and an increase in MAE from 1.7 dBm to 1.9 dBm. In addition, we evaluate the proposed FedLSTM considering variations in the number of participating clients and the number of local training epochs. The results show that even devices with limited computational power can meaningfully contribute to the training of the federated model, with fewer epochs achieving competitive results. Graphical analyses of the radio environment maps (REMs) generated by both FedLSTM and the centralized LSTM highlight their similarities. However, FedLSTM provides client privacy while reducing communication overhead and server strain.

Accurately identifying partial discharges (PDs) is crucial for ensuring the reliability of high-voltage switchgear components. This paper presents a novel framework that integrates advanced signal processing and deep learning techniques to enhance PD classification. Utilizing, logarithmic scaling and continuous wavelet transform (CWT), combined with an optimized convolutional neural network (CNN) for feature extraction, the framework outperforms traditional models in terms of classification accuracy and computational efficiency. Principal component analysis (PCA) is applied to further optimize the extracted features, reducing dimensionality and improving fault-specific feature selection. The proposed framework achieves a 100% classification accuracy, demonstrating superior performance when compared to existing models such as ResNet-50, DenseNet-201, MobileNetV2, and conventional deep CNNs. The framework not only achieves high accuracy but also significantly reduces training time, making it highly effective for real-time applications. With its ability to generalize across different PD types, the model provides a robust and computationally efficient solution for PD detection, offering significant advancements in the maintenance and reliability assessment of power system components.

Epilepsy is a prevalent neurological disorder with considerable risks, including physical impairment and irreversible brain damage from seizures. Given these challenges, the urgency for prompt and accurate seizure detection cannot be overstated. Traditionally, experts have relied on manual EEG signal analyses for seizure detection, which is labor-intensive and prone to hu-man error. Recognizing this limitation, the rise of deep learning methods has been heralded as a promising avenue, offering more refined diagnostic precision. On the other hand, the prevailing challenge in many models is their constrained emphasis on specific domains, potentially diminishing their robustness and precision in complex real-world environments. This paper presents a novel model that seamlessly integrates the salient features from the time-frequency domain alongside pivotal statistical attributes derived from EEG signals. This fusion process involves the integration of essential statistics, including the mean, median, and variance, combined with the rich data from compressed time-frequency (CWT) images processed using auto-encoders. This multidimensional feature set provides a robust foundation for subsequent analytic steps. A long short-term memory (LSTM) network, meticulously optimized for the renowned Bonn Epilepsy dataset, was used to enhance the capability of the proposed model. Preliminary evaluations underscore the prowess of the proposed model: remarkable 100% accuracy in most of the binary classifications, impressive performance exceeding 95% for three-class and four-class challenges, and a commendable rate exceeding 93.5% for the five-class classification.