检索结果-内蒙古大学图书馆

Respiratory Anomaly and Disease Detection using Multi-Level Temporal Convolutional Networks

学校读者我要写书评

暂无评论

IEEE Journal of Biomedical and Health Informatics 2025年 PP卷 PP页

作者： Le, Kim-Ngoc T. Byun, Gyurin Raza, Syed M. Le, Duc-Tai Choo, Hyunseung The Department of AI Systems Engineering Sungkyunkwan University Suwon Korea Republic of The James Watt School of Engineering University of Glasgow Glasgow United Kingdom The Department of Electrical and Computer Engineering Sungkyunkwan University Suwon Korea Republic of The Department of AI Systems Engineering Department of Electrical and Computer Engineering College of Computing and Informatics Sungkyunkwan University Suwon Korea Republic of

An automated analysis of respiratory sounds using Deep Learning (DL) plays a pivotal role in the early detection of lung diseases. However, current DL methods often examine the spatial and temporal characteristics of respiratory sounds in isolation, which inherently limit their potential. This study proposes a novel DL framework that captures spatial features through convolution operations and exploits the spatiotemporal correlations of these features using temporal convolution networks. The proposed framework incorporates MultiLevel Temporal Convolutional Networks (ML-TCN) to considerably enhance the model accuracy in detecting anomaly breathing cycles and respiratory recordings from lung sound audio. Moreover, a transfer learning technique is also employed to extract semantic features efficiently from limited and imbalanced data in this domain. Thorough experiments on the well-known ICBHI 2017 challenge dataset show that the proposed framework outperforms state-of-the-art methods in both binary and multi-class classification tasks for respiratory anomaly and disease detection. In particular, improvements of up to 2.29% and 2.27% in terms of the Score metric, average sensitivity and specificity, are demonstrated in binary and multi-class anomaly breathing cycle detection tasks, respectively. In respiratory recording classification tasks, the classification accuracy is improved by 2.69% for healthy-unhealthy binary classification and 1.47% for healthy, chronic, and non-chronic diagnosis These results highlight the marked advantage of the ML-TCN over existing techniques, showcasing its potential to drive future innovations in respiratory healthcare technology. © 2025 IEEE.

关键词： Lung cancer

Mitigation of Soiling and Assessment of PV Module Performance under IEC 60891 by Based-Brush and Based-Water Cleaning Methods Using

学校读者我要写书评

暂无评论

Applied Solar Energy (English translation of Geliotekhnika) 2024年第2期60卷 226-241页

作者： Zaoui, Fares Khenfer, Riad Lekbir, Abdelhak Mekhilef, Saad Rouabah, Zahir Materials and Electronic Systems Laboratory University of Mohamed El Bachir El Ibrahimi El- Anasser Bordj-Bou-Arreridj Algeria Department of Electrical Engineering University of Malaya Kuala Lumpur Malaysia School of Science Computing and Engineering Technologies Swinburne University of Technology Hawthorn VIC Australia Smart Materials and Renewables Energies Laboratory University of Mohamed El Bachir El Ibrahimi El-Anasser Bordj-Bou-Arreridj Algeria

Abstract: In the front surface of PV modules, the soiling and dust is a major issue, especially in areas with high soiling rates, frequent dust, limited water supplies, and significant solar energy potential and appears to have a significant influence in output power. This study proposes an approach to mitigate the soiling, dust and assess the PV module performance under IEC 60891 by using of based-brush and based-water cleaning methods (CBB and CBW), which can significantly increase power generation and reduce the cost of operation and maintenance. Moreover, the CBB and CBW cleaning types help to improve the performance of PV modules. The cleaning factor shows that the overall electrical output power is approximately 3.09%, with an average value of 1.61% in cleaning based-brush and 1.49% in cleaning-based water, the CBB improve the total output power with 52% and CBW with 58%. The cleaning process affects, positively by approximately 2.49 and 2.24%, and the maximum output current with an overall cleaning factor of approximately 3.58%, where the average cleaning factors for CBB and CBW were 1.9 and 1.68%, respectively. The CBB showed good performance and significantly reduced the thickness of the dust layer accumulated on the module surface and removed a large portion of soiling. Therefore, a regular wet or dry cleaning of PV modules surface is essentially needed and the combination between the two methods is important to attempt the minimum costs and maximum power. From a sustainability perspective, this work demonstrates that the CBB method can be significantly utilized to reduce soiling losses in PV modules without using water. The study shows that CBW is an effective way to remove the bird dropping, thin dust and improve the output power. © Allerton Press, Inc. 2024. ISSN 0003-701X, Applied Solar Energy, 2024, Vol. 60, No. 2, pp. 226–241. Allerton Press, Inc., 2024.

关键词： Solar panels

A Search Strategy and Vessel Detection in Maritime Environment Using Fixed-Wing UAVs

学校读者我要写书评

暂无评论

A Search Strategy and Vessel Detection in Maritime Environme...

International Symposium on Underwater Technology

作者： Marijana Peti Ana Milas Natko Kraševac Marko Križmančić Ivan Lončar Nikola Mišković Stjepan Bogdan Laboratory for Robotics and Intelligent Control Systems (LARICS) Faculty of Electrical Engineering and Computing University of Zagreb Zagreb Croatia Laboratory for Underwater Systems and Technologies (LABUST) Faculty of Electrical Engineering and Computing University of Zagreb Zagreb Croatia

In this paper, we address the problem of autonomous search and vessel detection in an unknown GNSS-denied maritime environment with fixed-wing UAVs. The main challenge in such environments with limited localization, communication range, and the total number of UAVs and sensors is to implement an appropriate search strategy so that a target vessel can be detected as soon as possible. Thus we present informed and non-informed methods used to search the environment. The informed method relies on an obtained probabilistic map, while the non-informed method navigates the UAVs along predefined paths computed with respect to the environment. The vessel detection method is trained on synthetic data collected in the simulator with data annotation tools. Comparative experiments in simulation have shown that our combination of sensors, search methods and a vessel detection algorithm leads to a successful search for the target vessel in such challenging environments.

关键词： Location awareness Navigation Annotations Computational modeling Search problems Probabilistic logic Sensors

QEdgeProxy: QoS-Aware Load Balancing for IoT Services in the computing Continuum

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Čilić, Ivan Jukanović, Valentin Žarko, Ivana Podnar Frangoudis, Pantelis Dustdar, Schahram Faculty of Electrical Engineering and Computing University of Zagreb Zagreb Croatia Distributed Systems Group TU Wien Vienna Austria

While various service orchestration aspects within computing Continuum (CC) systems have been extensively addressed, including service placement, replication, and scheduling, an open challenge lies in ensuring uninterrupted data delivery from IoT devices to running service instances in this dynamic environment, while adhering to specific Quality of Service (QoS) requirements and balancing the load on service instances. To address this challenge, we introduce QEdgeProxy, an adaptive and QoS-aware load balancing framework specifically designed for routing client requests to appropriate IoT service instances in the CC. QEdgeProxy integrates naturally within Kubernetes, adapts to changes in dynamic environments, and manages to seamlessly deliver data to IoT service instances while consistently meeting QoS requirements and effectively distributing load across them. This is verified by extensive experiments over a realistic K3s cluster with instance failures and network variability, where QEdgeProxy outperforms both Kubernetes built-in mechanisms and a state-of-the-art solution, while introducing minimal computational overhead. © 2024, CC BY.

关键词： Quality of service

Comparative Analysis of Programming by Demonstration Methods: Kinesthetic Teaching vs Human Demonstration

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Maric, Bruno Zoric, Filip Petric, Frano Orsag, Matko The LARICS Laboratory for Robotics and Intelligent Control Systems University of Zagreb Faculty of Electrical Engineering and Computing Zagreb10000 Croatia

Programming by demonstration (PbD) is a simple and efficient way to program robots without explicit robot programming. PbD enables unskilled operators to easily demonstrate and guide different robots to execute task. In this paper we present comparison of demonstration methods with comprehensive user study. Each participant had to demonstrate drawing simple pattern with human demonstration using virtual marker and kinesthetic teaching with robot manipulator. To evaluate differences between demonstration methods, we conducted user study with 24 participants which filled out NASA raw task load index (rTLX) and system usability scale (SUS). We also evaluated similarity of the executed trajectories to measure difference between demonstrated and ideal trajectory. We concluded study with finding that human demonstration using a virtual marker is on average 8 times faster, superior in terms of quality and imposes 2 times less overall workload than kinesthetic teaching. © 2024, CC BY.

关键词： Demonstrations

Tri-Band Endfire Antenna Array with Wide Angles of Beam-Scanning Capability for 5G mmWave Mobile Communication

学校读者我要写书评

暂无评论

Tri-Band Endfire Antenna Array with Wide Angles of Beam-Scan...

2023 IEEE International Workshop on Technologies for Defense and Security, TechDefense 2023

作者： Zidour, Ali Ayad, Mouloud Alibakhshikenari, Mohammad Basherlou, Haleh Jahanbakhsh Parchin, Naser Ojaroudi See, Chan Hwang University of m'Hamed Bougara Faculty of Technology Department of Electrical Systems Engineering Boumerdes Algeria University of Bouira Faculty of Sciences and Applied Sciences Department of Electrical Systems Bouira Algeria Universidad Carlos Iii de Madrid Leganés Department of Signal Theory and Communications Madrid Spain Edinburgh Napier University School of Computing Engineering and the Built Environment Edinburgh United Kingdom

ISBN: (纸本)9798350319392

In this paper, a novel multiband phased array antenna with endfire radiation pattern is designed at millimeter-wave (mmWave) frequency bands for 5G mobile applications. The proposed antenna element is a compact quasi-Yagi with a modified dipole driver to generate multiband operation, arranged on low-loss RO4003C materiel. The antenna array achieves excellent performance and covers wide frequency ranges of (24.15-29.8 GHz), (36.2-50 GHz) to support 5G NR bands;26, 28, 39 and the new allocated 48 GHz. The array can exhibit a maximum realized gain from 8.4 dBi to 11.8 dBi. Meanwhile, wide beam-scanning angles from - 68° to 65° at 28 GHz, from 40° to 41° at 39 GHz, and from - 33° to 35° at 48 GHz are realized. © 2023 IEEE.

关键词： Millimeter waves

Inference Load-Aware Orchestration for Hierarchical Federated Learning

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Lackinger, Anna Frangoudis, Pantelis A. Čilić, Ivan Furutanpey, Alireza Murturi, Ilir Žarko, Ivana Podnar Dustdar, Schahram Distributed Systems Group TU Wien Vienna Austria Faculty of Electrical Engineering and Computing University of Zagreb Zagreb Croatia

Hierarchical federated learning (HFL) designs introduce intermediate aggregator nodes between clients and the global federated learning server in order to reduce communication costs and distribute server load. One side effect is that machine learning model replication at scale comes "for free" as part of the HFL process: model replicas are hosted at the client end, intermediate nodes, and the global server level and are readily available for serving inference requests. This creates opportunities for efficient model serving but simultaneously couples the training and serving processes and calls for their joint orchestration. This is particularly important for continual learning, where serving a model while (re)training it periodically, upon specific triggers, or continuously, takes place over shared infrastructure spanning the computing continuum. Consequently, training and inference workloads can interfere with detrimental effects on performance. To address this issue, we propose an inference load-aware HFL orchestration scheme, which makes informed decisions on HFL configuration, considering knowledge about inference workloads and the respective processing capacity. Applying our scheme to a continual learning use case in the transportation domain, we demonstrate that by optimizing aggregator node placement and device-aggregator association, significant inference latency savings can be achieved while communication costs are drastically reduced compared to flat centralized federated learning. © 2024, CC BY.

关键词： Federated learning

Recent Developments in Authentication Schemes Used in Machine-Type Communication Devices in Machine-to-Machine Communication: Issues and Challenges

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2024年第4期79卷 93-115页

作者： Shafi Ullah Sibghat Ullah Bazai Mohammad Imran Qazi Mudassar Ilyas Abid Mehmood Muhammad Asim Saleem Muhmmad Aasim Rafique Arsalan Haider Ilyas Khan Sajid Iqbal Yonis Gulzar Kauser Hameed Department of Computer Engineering Balochistan University of Information TechnologyEngineering and Management SciencesQuetta87300Pakistan Department of Information Technology Balochistan University of Information TechnologyEngineeringand Management SciencesQuetta87300Pakistan Department of Information Systems College of Computer Sciences and Information TechnologyKing Faisal UniversityHofuf in Al-Ahsa31982Saudi Arabia Department of Management Information Systems College of Business AdministrationKing Faisal UniversityHofuf in Al-Ahsa31982Saudi Arabia Department of Software Engineering College of ComputingRiphah International UniversityFaisalabad44000Pakistan Department of Electrical Engineering Balochistan University of Information Technology Engineeringand Management SciencesQuetta87300Pakistan Department of Mathematics College of Science Al-ZulfiMajmaah UniversityAl-Majmaah11952Saudi Arabia

Machine-to-machine (M2M) communication plays a fundamental role in autonomous IoT (Internet of Things)-based infrastructure, a vital part of the fourth industrial revolution. Machine-type communication devices(MTCDs) regularly share extensive data without human intervention while making all types of decisions. Thesedecisions may involve controlling sensitive ventilation systems maintaining uniform temperature, live heartbeatmonitoring, and several different alert systems. Many of these devices simultaneously share data to form anautomated system. The data shared between machine-type communication devices (MTCDs) is prone to risk dueto limited computational power, internal memory, and energy capacity. Therefore, securing the data and devicesbecomes challenging due to factors such as dynamic operational environments, remoteness, harsh conditions,and areas where human physical access is difficult. One of the crucial parts of securing MTCDs and data isauthentication, where each devicemust be verified before data transmission. SeveralM2Mauthentication schemeshave been proposed in the literature, however, the literature lacks a comprehensive overview of current M2Mauthentication techniques and the challenges associated with them. To utilize a suitable authentication schemefor specific scenarios, it is important to understand the challenges associated with it. Therefore, this article fillsthis gap by reviewing the state-of-the-art research on authentication schemes in MTCDs specifically concerningapplication categories, security provisions, and performance efficiency.

关键词： Authentication cyber security internet of things machine-type communication devices machine-to-machine communication

An Energy-Efficient Artefact Detection Accelerator on FPGAs for Hyper-Spectral Satellite Imagery

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Castelino, Cornell Khandelwal, Shashwat Shreejith, Shanker Bogaraju, Sharatchandra Varma Reconfigurable Computing Systems Lab Electronic & Electrical Engineering Trinity College Dublin Ireland Faculty of Computing Ulster University Jordanstown Campus Newtownabbey United Kingdom

Hyper-Spectral Imaging (HSI) is a crucial technique used to analyse remote sensing data acquired from Earth observation satellites. The rich spatial and spectral information obtained through HSI allows for better characterisation and exploration of the Earth’s surface over traditional techniques like RGB and Multi-Spectral imaging on the downlinked image data at ground stations. In some cases, these images do not contain meaningful information due to the presence of clouds or other artefacts, limiting their usefulness. Transmission of such artefact HSI images leads to wasteful use of already scarce energy and time costs required for communication. While detecting such artefacts prior to transmitting the HSI image is desirable, the computational complexity of these algorithms and the limited power budget on satellites (especially CubeSats) are key constraints. This paper presents an unsupervised learning-based convolutional autoencoder (CAE) model for artefact identification of acquired HSI images at the satellite and a deployment architecture on AMD’s Zynq Ultrascale FPGAs. The model is trained and tested on widely used HSI image datasets: Indian Pines, Salinas Valley, the University of Pavia and the Kennedy Space Center. For deployment, the model is quantised to 8-bit precision, fine-tuned using the Vitis-AI framework and integrated as a subordinate accelerator using AMD’s Deep-Learning Processing Units (DPU) instance on the Zynq device. Our tests show that the model can process each spectral band in an HSI image in 4 ms, 2.6× better than INT8 inference on Nvidia’s Jetson platform & 1.27× better than SOTA artefact detectors. Our model also achieves an f1-score of 92.8% and FPR of 0% across the dataset, while consuming 21.52 mJ per HSI image, 3.6× better than INT8 Jetson inference & 7.5× better than SOTA artefact detectors, making it a viable architecture for deployment in CubeSats. © 2024, CC BY.

关键词： Deep learning