This paper introduces a methodology for estimating mayfly populations by utilizing radar data from the US National Weather Service's NEXRAD system. The research emphasizes the integration of Python-based tools, including PyEphem, Boto3, and Py-ART, to efficiently retrieve, process, and filter radar data, thereby improving the accuracy of mayfly emergence monitoring. The filtration process leverages advanced polarimetric radar techniques, allowing for the effective distinction between meteorological and biological targets, such as insects. The method estimates mayfly population density by calculating the total backscattering area from radar reflectivity data, which is then used to derive the number of mayflies in the surveyed region. This approach provides a robust and reliable technique for ecological monitoring and can be applied to studies of insect migrations and environmental shifts within the Upper Mississippi River Region (UMRR).
Smart motor controllers usually provide plentiful information that makes it possible to model the various mechanisms they control. One such mechanism, a flywheel, is one of the simplest models to implement, yet has various characteristics that need to be identified for the model to be accurate. To ensure accurate control of the flywheel, not only is real-time state information necessary such as position, velocity, and acceleration, but also system characteristics such as inertia, friction, and drag. This paper investigates the use of extended Kalman filter to simultaneously estimate the system state and parameters, providing rich information that the user could use to optimally control the end mechanism.
As a social media platform, Twitter significantly influences stock market movement in the contemporary digital era. Twitter users can share real-time events and opinions about financial news instantly among the mass population. However, tweets sentiments on financial news have complexity considering sentiment ambiguity, financial jargon, and lack of context and linguistic style. Here, we employed machine learning models (Random Forest, XGBoostGradient Boosting, support vector Machine, Logistic Regression), pre-trained models (RoBERTa, DistilBERT, FinBERT, XLNet, and Vedar), and deep learning models (LSTM, GRU, and a combination of CNN and LSTM) to analyze stock market sentiment utilizing the Twitter dataset. We also deployed a word cloud analysis to identify investors' sentiments. This research utilized a Twitter stock news dataset[7] [binary sentiment labeled as positive (1) or negative (-1)] from Kaggle containing 5971 tweets. Furthermore, we reported the effectiveness of the multiple computational methods in classifying financial sentiment. The study findings demonstrated that DistilBERT, the pre-trained model, outperformed all the machine learning, deep learning, and other pre-trained models. It had the highest accuracy score of 80%, the highest F-1 score of 84.10%, and RoBERTa had the highest area under the curve score of 86.43%. As per word cloud analysis, market sentiment of tweets are well represented by the resulting cloud word (e.g, “short”, “volume”, “low”, “bullish”, “earning”, “target”, “breakout”). This research benefits investors, financial analysts, AI researchers, and hedge funds by providing a robust sentiment-driven stock price prediction framework. Future work should focus on multi-source sentiment incorporation and enhance domain-specific knowledge to enhance predictive accuracy and reliability.
Cloud services are helping with the management and organizational placement of organizational IT resources by providing products that are malleable, versatile, and inexpensive. But as people start using cloud-based applications in areas that cannot afford any disruption in healthcare, finance, public safety and defense, the need for sturdy and reliable systems becomes critical. For critical business processes, performance disruption, availability problem or failure conditions, risk exposure that causes loss, brand deterioration, or death is not acceptable. Durability architectures are built to avoid, accommodate and restore disruptions resulting from equipment breakdown, cyber- attacks, calamities or surge in workload. This paper focuses on fault tolerance, replication of data, disaster recovery, and distributed systems, as methods that can be used to construct reliable systems. It also explores newer trends like Artificial Intelligence, Machine Learning, and Edge Computing that allow the continuous monitoring, analytics, and automated recovery. The issues are presented and discussed in detail, and the trends that include on-premise and cloud environments, compliance issues and evolving threat landscape are highlighted. Specific steps to increase business productivity when using business intelligence, credibility of information, and adaptability are described. This work positions resilience as a competitive advantage underpinning the capabilities of cloud-oriented organisations to navigate the challenges of current technology environments cohesively.
Credit cards are a type of modern financial asset that is convenient, secure, and beloved by consumers. However, a huge amount of fraud in financial transactions is occurring on a daily basis, so it is important to have effective fraud detection systems to protect digital payment. Traditional rule based systems for fraud detection are however not very effective in dynamic environments as they cannot detect new patterns and methods used by fraudsters. Addressing this, this study explores four machine learning models, including Graph Neural Networks (GNN), Long Short Term Memory (LSTM), Multilayer Perceptron (MLP), and Transformer to detect credit card fraud in an imbalanced dataset. For use of the synthetic minority over-sampling technique (SMOTE) in the imbalanced dataset, the machine learning models demonstrate a great improvement in classification accuracy. The results reveals the effectiveness of each model, with GNNs and LSTMs being the best in managing the complexities of transaction data.
The increasing demand for reliable, high-speed communication in 5G applications highlights the need for highperformance antennas in Wireless Body Area Networks (WBANs). Traditional single-element microstrip antennas face limitations in gain and bandwidth, especially when placed near the human body, where electromagnetic absorption reduces performance. This study addresses these challenges by designing a circular microstrip patch antenna and evaluating array configurations (1×2, 1×4, 2×2) to optimize key performance metrics such as gain, bandwidth, and efficiency. The target attribute is to enhance gain and bandwidth at 39 GHz, a frequency selected for its suitability in mmWave 5G applications, offering high data rates and reduced interference. The antenna, built on a Rogers RT5880 substrate with a 1.57 mm thickness, achieved a gain of 7.385 dBi and a bandwidth of 6.4 GHz in free-space simulations. On-body simulations with a human torso phantom showed significant improvements for array configurations: the 1×4 array reached a peak gain of 9.73 dBi, a bandwidth of 8.18 GHz, and a reflection coefficient of 49.69 dB at a 4 mm distance. The 2×2 array provided a gain of 9.23 dBi and a radiation efficiency of 81.28%. These results demonstrate that the array antennas, particularly the 1×4 and 2×2 configurations, outperform single-element designs, making them ideal for high-frequency WBAN applications in the emerging 5G landscape.
The demand for efficient and compact inverter solutions in electric drive systems, particularly in electric aircraft and other high-frequency applications, has prompted significant interest in multilevel inverter (MLI) topologies. Conventional MLIs are constrained by high switching losses and the complexity of modulation strategies, especially at fundamental frequencies above 2 kHz, the pulse-width modulation (PWM) techniques may become inefficient, as they require carrier frequencies nearly ten times the fundamental frequency or higher, significantly increasing switching losses in the converter. To overcome these challenges, this paper presents a new seven-level twelve-step inverter (SLTSI) that eliminates the need for high-frequency carrier-based PWM. The proposed inverter utilizes three planar high-frequency transformers and twelve switches to deliver high performance with a simplified control structure. Operating at the fundamental frequency, the SLTSI topology improves DC source utilization and achieves a peak phase voltage of 4 $V_{dc}$ /3, leading to an enhanced levels-to-source ratio of 7/1. Comprehensive simulations and experimental measurements validate the inverter's performance and demonstrate its suitability for compact, high-speed motor drive systems.
Agentic AI Systems represent a significant ad-vancement in artificial intelligence by enabling systems to autonomously perceive, decide, and act in complex environments. This review explores the definition, scope, advantages, challenges, opportunities, and trustworthiness of agentic AI in organizational and societal contexts. Agentic AI offers higher efficiency, scalability, and improved decision making which enables the organizations to streamline operations and enhance their productivity. Despite these advancements Agentic AI has its own challenges such as; safety concerns, accountability, reliability issues and potential misuse remain critical areas for consideration. This paper provides a comprehensive discussion of agentic AI's impact in emphasizing both its transformative potential and the need for continuous oversight and refinement.
The advent of quantum computing will revolutionize computation wherever commercial operations depend on industrial automation. Data center operations will be impacted by the approach. Considering this, the brief analysis of how quantum computing can upend the data center sector that follows reflects Classical bits are used in classical computers. There are just two states that may be characterized as classical bits: zero and one. In quantum computing, qubits—also referred to as quantum bits—are used. Qubits may exist in several states at once. This characteristic is known as superposition. The most basic element of quantum computing is this. The second fundamental idea in quantum computing is entanglement. No matter how far away two or more qubits are, entanglement may bring them together. One qubit's state immediately impacts the others when they are entangled. The foundation of quantum parallelism is superposition and entanglement. Quantum algorithms may examine several possible answers at once thanks to quantum parallelism. This is especially helpful for addressing problems since traditional computers would need to analyze every event in detail. Computer speed is greatly increased by quantum parallelism. It is thus very helpful for handling challenging tasks like cryptography. In manufacturing sites specially into battery cell manufacturing segment, and quantum-edge computing plays a vital role while considering a data center setup to support their business in day-to-day operational. In manufacturing sites, especially in the battery cell manufacturing segment, quantum-edge computing plays a vital role while considering a data center setup to support their business in day-to-day operations. In this paper, we will be sharing the potential advantages of quantum computing and edge computing in a datacenter that is engaged to support battery cell manufacturing.
This paper presents the design and implementation of BrightPath, an IoT-enabled smart bike system aimed at enhancing cyclist safety and navigation, particularly during nighttime rides. The system integrates real-time navigation, dynamic turn signaling, and LED-based safety boundary projection to improve visibility and communication between cyclists and motorists. The proposed system demonstrates novel embedded system integration by coordinating real-time communication between micro controllers, dynamic sensor-based decision-making, and sustainable power management. Unlike existing solutions, BrightPath features an automated signaling and navigation interface managed by tightly coupled embedded software on both the Arduino and Raspberry Pi platforms. These contributions highlight innovations in multi-controller synchronization, sensor fusion for real-time motion detection, and self-sustaining energy-aware embedded processing.
This paper presents a method for thermal image recognition utilizing the YOLO v11 model deployed on Jetson Orin Nano. The system utilizes the Tensor RT cores to accelerate the recognition task, enhancing performance and power efficiency. Thermal image data for the recognition are from Teledyne FLIR Free ADAS (advanced driver assistance system) Thermal Dataset v2. The model is trained on the YOLO-v11 model, which achieves 71% precision scores on a small-size model and 59% of mAP50 (mean Average Precision at 50% Intersection over Union (IoU) for object detection) accuracy scores on a medium-size model. Results show Jetson Orin Nano can perform approximately 36.5 frames per second (fps) on a small-size model with 6 watts of total module power consumption or 1.9 watts on CPU and GPU cores. The YOLO v11 model performs approximately 15% better than the YOLO v8.
With the advances in technology enhancing our ability to record abdominal sounds, the study and analysis of bowel sounds or phonoenterography have drawn significant attention. Clinicians and researchers have attempted to explore the variations in bowel sounds in structural, functional, and neurodegenerative conditions as well as guide the timely initiation of feeds and prediction of complications like paralytic ileus in postsurgical patients. Despite these applications, limited efforts have been made to understand the normal physiology of bowel sounds and their variation with meals. This pilot feasibility study investigates the trends in bowel sounds with different meal compositions at two locations in the abdomen on a healthy subject. The authors also define 4 metrics for phonoenterogram analytics, namely, bowel rate, average bowel rate, average bowel rate variability, and the phonoenterogram (PEG) index. The PEG index was found to be consistent irrespective of the meal constituents or location of the recording. Larger studies are needed to validate and generalize these findings in healthy subjects. Further studies in patients with various gastrointestinal diseases can potentially lead to the development of a novel, noninvasive AI powered diagnostic modality which can transform clinical GI practice.
Distributed machine learning offers significant advantages over centralized approaches, including improved accessibility, memory efficiency, and performance. However, despite these benefits, distributed training remains underused, often sacrificed in favor of data security-centric methods such as federated learning. A major barrier to wider adoption is the lack of effective algorithms that efficiently leverage distributed compute. Current training methods often rely on bandwidth-heavy all-gather operations and global gradient dependencies, making distributed training challenging. In this paper, we propose a novel approach to distributing layer-wise knowledge distillation. By carefully routing data, passing layer parameters instead of activations, and eliminating cross-device gradient dependencies, our method significantly reduces training wall time compared to data-parallelization. Furthermore, our approach reduces communication bandwidth, improves memory utilization, and enables flexible resource allocation.
This research investigates the optimization of Convolutional and Dense Neural Networks (CNNs and DNNs) for autonomous steering using the (N + M) Evolution Strategy (ES) with the 1/5th success rule. The primary objective is to develop a lightweight CNN based model capable of real-time steering angle prediction, mimicking human driving behavior on predefined paths. The ES algorithm automates hyperparameter tuning, dynamically adjusting parameters such as filter sizes and layer configurations. Data collection encompasses driving scenarios recorded via the LTU ACTor autonomous driving platform, including variations in path direction and driving style. The very small dataset consists of timestamped images labeled with steering angles and pre-processed to focus on relevant visual information. Initial experiments involve training a baseline CNN model, which is then refined using ES to significantly reduce the size of the model while maintaining competitive predictive accuracy. The results highlight the viability of lightweight neural network architectures for real-time autonomous systems, striking a balance between computational efficiency and performance. This study not only advances research initiatives on the use of evolutionary algorithms for autonomous driving applications but also lays the foundation for the deployment of cost-effective and scalable solutions in self-driving technology.
Network Intrusion Detection Systems (NIDS) play a critical role in cybersecurity but face challenges such as high-dimensional data, feature redundancy, and class imbalance. Traditional feature selection methods often struggle with efficiency and accuracy in these complex scenarios. This paper proposes a Quantum Annealing (QA) approach to address the challenges of the feature engineering in NIDS by maximizing relevance, minimizing redundancy, and reducing computational overhead. Using the UNSW-NB15 dataset, which reflects real-world network traffic, we formulate the feature selection problem as a Quadratic Unconstrained Binary Optimization (QUBO) and solve it using D-Wave's Quantum Annealer. The selected features are evaluated with Random Forest (RF) and Multilayer Perceptron (MLP) classifiers. Simulation results show that QA reduces feature selection time by 60% compared to wrapper methods, achieving 99.14% accuracy with RF and 98.87% with MLP, while significantly improving detection rates for minority attack classes. Unlike conventional methods, QA effectively addresses imbalanced datasets without requiring artificial balancing, making it well-suited for real-world NIDS applications. This work demonstrates the potential of quantum-assisted cybersecurity, paving the way for scalable and high-performance intrusion detection systems.
Falls are a leading cause of hospitalization among seniors, with delayed medical response and underreporting being the most critical factors. RiskWatch is a wearable fall detection and health-monitoring device built to address this critical gap in senior healthcare. Our solution utilizes an ESP32-based wearable equipped with an inertial measurement unit (IMU), a pulse sensor for vital sign tracking, and a lightweight Machine Learning model that classifies incoming sensor data to distinguish fall events from normal activities. When the system detects a fall, the device instantly transmits data to a companion Android application via Bluetooth Low Energy (BLE), which in turn updates a secure, cloud-based database. In preliminary tests involving simulated fall scenarios, the prototype demonstrated reliable, real-time detection and seamless caregiver notification, thereby significantly reducing medical response latency. Furthermore, the device captures contextual data-such as heart rate and incident timestamp-to enable post-fall analysis and help identify patterns that can inform preventative measures. By integrating immediate alerts, data storage, and caregiver access in a single system, RiskWatch offers a proactive solution to reduce fall-related risks and enhance accountability in healthcare settings.
The rapid advancement of emerging technologies has significantly impacted the concept of digital sovereignty, presenting both opportunities and challenges for nations trying to control their digital ecosystems. This paper examines the relationship between digital sovereignty and cybersecurity, focusing on how technologies like Artificial Intelligence, Blockchain, and 5G are reshaping digital sovereignty and influencing national policies. It highlights both the transformative potential of these technologies and the associated risks. The paper also reviews global governance and collaborative cybersecurity frameworks in digital sovereignty, exploring how international cooperation can strengthen cybersecurity efforts. By analyzing different national approaches, the study highlights the need for governments and organizations to balance robust cybersecurity with maintaining control over their digital territories.
Epilepsy is one of the most common neurological disorders and characterized by unpredictable recurrent seizures that affect people of different ages. Certain seizures can result in severe, uncontrollable bodily motions that injure patients or cause Sudden Unexpected Death in Epilepsy (SUDEP). Help is desperately needed because the majority of these symptoms are associated with generalized tonic-clonic seizures (G TCS). Surface electromyography (sEMG) has been clinically verified for automated GTCS identification and can be utilized for non-invasive G TCS detection. However, classification accuracy and computing efficiency are challenged by the high-dimensional and complicated nature of EMG data. In order to improve GTCS's classification performance with EMG signals, this research inves-tigates feature optimization strategies. Various feature extraction methods are applied in the signal-specific and signal independent domain, followed by ANOVA technique. After extracting 192 features in total for both types, we ranked the top 10 features in both types that performed optimally and could be crucial in GTCS classification. The effectiveness of these optimized features is evaluated using machine learning models. We classified the dataset by using the Random Forest model (95%), K-nearest Neighbor Algorithm (90%), Extreme Gradient Boost (95%), Decision Tree (93 %), Artificial Neural Network (97 %) and Long-short Term Memory (96%) and compared the results with one another for the G TCS classification. Experimental results demon-strate that optimized feature selection significantly improves classification accuracy while reducing computational complexity. This study highlights the importance of feature optimization in EMG signal for GTCS classification and understanding the severity of the seizures in patients.
With the increasing adoption of various sensors, human action recognition has gained significant attention across multiple domains, including person surveillance and human-robot interaction. However, existing data-driven approaches struggle with effectively modeling the spatiotemporal dynamics of sensory data and suffer from limited generalization capability. To address these challenges, this paper introduces a novel graph-based deep learning framework, incorporating a Graph-Attentive Variational Sparse Contractive Peephole LSTM (GAVSC-PLSTM) model. The proposed architecture effectively captures spatiotemporal correlations among sensory data from different body parts and introduces a novel encoder-decoder generative framework to extract task-relevant deep spatiotemporal features. Extensive experiments on three widely used public datasets demonstrate that the proposed model outperforms recent baseline methods.
The paper discussed the introduction of a YOLO-based AI system for the detection of road incidents in real-time as an effort toward enhanced road safety by automating the detection and response to road-related risks such as accidents and speeding vehicles. Using YOLOv8 and YOLO11x models trained on a broad dataset of 15,000 traffic scenario images, the system detects various types of accidents and vehicle speeds with high accuracy and efficiency under different conditions. The proposed solution addresses main challenges of existing surveillance systems through real-time processing, sensor-free speed estimation, and robustness against low-resolution inputs, thus proving to be a great promise for improving traffic management and safety.
