A new light ensemble deep-learning framework to detect fire DOI Open Access
Ahmed A. Alsheikhy, Tawfeeq Shawly,

Hossam Ahmed

et al.

International Journal on Information Technologies and Security, Journal Year: 2023, Volume and Issue: 15(4), P. 37 - 48

Published: Nov. 30, 2023

Fires can cause devastating damage to lands, properties, and humans. Many countries suffer from huge financial losses due these fires. Therefore, there is a need implement practical solution spot fires effectively accurately. Deep-learning algorithms artificial intelligence have been deployed recently in various fields, such as monitoring systems, economics, detection. This paper proposes New Light Ensemble Deep-Learning Framework (NLEDLF). framework consists of two deep-learning technologies, which are Generative Adversarial Network (NGAN) Convolutional Neural (NCNN). These tools incorporated into the along with some image preprocessing methods detect using pixels. The proposed achieves reasonable.

Language: Английский

A Survey of Internet of Things and Cyber-Physical Systems: Standards, Algorithms, Applications, Security, Challenges, and Future Directions DOI Creative Commons
Kwok Tai Chui, Brij B. Gupta, Jiaqi Liu

et al.

Information, Journal Year: 2023, Volume and Issue: 14(7), P. 388 - 388

Published: July 8, 2023

The smart city vision has driven the rapid development and advancement of interconnected technologies using Internet Things (IoT) cyber-physical systems (CPS). In this paper, various aspects IoT CPS in recent years (from 2013 to May 2023) are surveyed. It first begins with industry standards which ensure cost-effective solutions interoperability. With ever-growing big data, tremendous undiscovered knowledge can be mined transformed into useful applications. Machine learning algorithms taking lead achieve target applications formulations such as classification, clustering, regression, prediction, anomaly detection. Notably, attention shifted from traditional machine advanced algorithms, including deep learning, transfer data generation provide more accurate models. years, there been an increasing need for security techniques defense strategies detect prevent being attacked. Research challenges future directions summarized. We hope that researchers conduct studies on CPS.

Language: Английский

Citations

24
Digital Twin Heuristic Positioning of Insulation in Multimodal Electric Systems DOI
Andrzej Sikora, Adam� Zielonka, Muhammad Fazal Ijaz

et al.

IEEE Transactions on Consumer Electronics, Journal Year: 2024, Volume and Issue: 70(1), P. 3436 - 3445

Published: Feb. 1, 2024

We present a diagnostic method which uses fuzzy voltage wave to test the insulation systems. Proposed solution can be used in diagnostics of electrical machines and devices. Our allows determine parameters elements an diagram. For this purpose, mathematical model coil system is built evolutionary optimization employing Red Fox Optimization algorithm proposed given specific industrial setting. The process carried out parallel mode, makes feasible for real-world applications. Research experiments show high efficiency various scenarios.

Language: Английский

Citations

8
Advancing Healthcare and Elderly Activity Recognition: Active Machine and Deep Learning for Fine- Grained Heterogeneity Activity Recognition DOI Creative Commons
Sidra Abbas, Gabriel Avelino Sampedro, Shtwai Alsubai

et al.

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 44949 - 44959

Published: Jan. 1, 2024

This research explores how technology can be used to understand and identify activities among elderly individuals. By utilizing HAR70+ data applying methods like Active Learning (AL), Machine (ML), Deep (DL), this aims predict various performed by older adults. Moreover, the study leverages dataset, providing insight into daily of individuals AL-based ML DL techniques construct predictive models for these activities. The experiments are presented systematically, summarizing outcomes machine-learning across three iterative experiments. explored a diverse array algorithms, including Random Forest (RF), Extreme Gradient Boosting (XGBoost), Logistic Regression (LR), K-Nearest Neighbors (KNN), Stochastic Descent (SGB) such as Neural Networks (DNN) Long Short-Term Memory networks (LSTM) experimentation. trained on 7 activities: walking, shuffling, climbing stairs (up down), standing, sitting, lying down, 4 separately: using same method. Results reveal that LSTM achieved best accuracy 0.98 0.95 RF actives, showing potential techniques, particularly when integrated with AL, enhance activity recognition rate, patient care, optimize medication strategies improve well-being

Language: Английский

Citations

5
Hybrid System for Intelligent Context Situation Detection DOI

Ikhlass Mastour,

Hela Zorgati, Raoudha Ben Djemaa

et al.

Communications in computer and information science, Journal Year: 2025, Volume and Issue: unknown, P. 460 - 472

Published: Jan. 1, 2025

Language: Английский

Citations

0
MCT-CNN-LSTM: A Driver Behavior Wireless Perception Method Based on an Improved Multi-Scale Domain-Adversarial Neural Network DOI Creative Commons
K.M. Chen, Yingxue Diao, Yucheng Wang

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(7), P. 2268 - 2268

Published: April 3, 2025

Driving behavior recognition based on Frequency-Modulated Continuous-Wave (FMCW) radar systems has become a widely adopted paradigm. Numerous methods have been developed to accurately identify driving behaviors. Recently, deep learning gained significant attention in signal processing due its ability eliminate the need for intricate preprocessing and automatic feature extraction capabilities. In this article, we present network that incorporates multi-scale channel-time modules, referred as MCT-CNN-LSTM. Initially, multi-channel convolutional neural (CNN) combined with Long Short-Term Memory Network (LSTM) is employed. This model captures both spatial features temporal dependencies from input signal. Subsequently, an Efficient Channel Attention (ECA) module utilized allocate adaptive weights channels carry most relevant information. final step, domain-adversarial training applied extract common source target domains, which helps reduce domain shift. approach enables accurate classification of behaviors by effectively bridging gap between domains. Evaluation results show our method reached accuracy 97.3% real measured dataset.

Language: Английский

Citations

0
Exploring the impact of digital twin technology in infrastructure management: a comprehensive review DOI Creative Commons
Shi Qiu, Qasim Zaheer,

Fahad Ali

et al.

Journal of Civil Engineering and Management, Journal Year: 2025, Volume and Issue: 31(4), P. 395 - 417

Published: April 29, 2025

This paper examines the role of Digital Twin Technology (DTT) in transforming infrastructure management, with a focus on sustainability. It highlights how advancements Artificial Intelligence (AI), Building Information Modeling (BIM), and Internet Things (IoT) are driving effectiveness Twins real-world applications. Through detailed case studies, showcases practical benefits DTT across various sectors. also evaluates current trends strategies for enhancing integration into systems. The research reveals striking 80% increase DTT-related publications from 2019 to 2024, Asia, particularly China, leading contributions. concludes by addressing future potential, challenges, risks DTT, offering valuable insights stakeholders aiming optimize management digital era.

Language: Английский

Citations

0
Augmenting context with power information for green context-awareness in smart environments DOI Creative Commons
Umar Mahmud, Shariq Hussain

Frontiers in Computer Science, Journal Year: 2024, Volume and Issue: 6

Published: March 7, 2024

The increase in the use of smart devices has led to realization Internet Everything (IoE). heart an IoE environment is a Context-Aware System that facilitates service discovery, delivery, and adaptation based on context classification. been defined domain-dependent way, traditionally. classical models have focused rich lack Cost Context (CoC) can be used for decision support. authors present philosophy-inspired mathematical model includes confidence activity classification context, actions performed, power information. Since single recurring lead distinct performed at different times, it better record actions. information consumed complete processing quality attribute context. Power consumption useful metric as CoC suitable power-constrained awareness. To demonstrate effectiveness proposed work, example contexts are described, presented mathematically this study. aggregated with information, outcome concept situational results show gathered through sensor data deduced remote services made more parameters.

Language: Английский

Citations

1
A novel wavelet energy feature for damage identification with a digital twin considering measurement uncertainties DOI Creative Commons

Xiaobang Zhang,

Yong Lu

Journal of Structural Integrity and Maintenance, Journal Year: 2024, Volume and Issue: 10(1)

Published: Dec. 19, 2024

In structural health monitoring and damage identification literature, supervised machine learning has been commonly adopted. However, the establishment of training dataset remains to be an open question. Besides indirect experimental methods such as adding masses, use a digital replica (digital twin) in reference, undamaged state is deemed necessity, so that variety future damaged states may generated by varying properties twin. little research available literature addresses challenge modelling errors approach. This study advances digital-twin-based approach examining ability twin generate wavelet packet node energy (WPNE) features for identifying influences inherent uncertain physical properties, particularly damping. A novel WPNE feature developed through engineering, effectively mitigating inaccuracies brought about damping estimates. The proposed with new validated via numerical laboratory experiments, demonstrating its robustness against inevitable errors. work brings role twins step further towards real-life applications.

Language: Английский

Citations

1
A Bilateral Assessment of Human Activities Using PSO-Based Feature Optimization and Non-linear Multi-task Least Squares Twin Support Vector Machine DOI
Ujwala Thakur, Ankit Vidyarthi, Amarjeet Prajapati

et al.

SN Computer Science, Journal Year: 2024, Volume and Issue: 5(3)

Published: March 13, 2024

Language: Английский

Citations

0
Sensor-Based Human Activity Recognition Using a Hybrid CNN-SVM Approach DOI

Imene Charabi,

M’hamed Bilal Abidine, Belkacem Fergani

et al.

Published: April 21, 2024

Human Activity Recognition (HAR) is the process of interpreting human actions from sensor data. This paper presents a hybrid approach for HAR utilizing Convolutional Neural Network (CNN) feature extraction and Support Vector Machine (SVM) classification. The model end-to-end trainable, where SVM classifier replaces softmax layer CNN. Evaluation was conducted on two benchmark datasets, UCI UniMiB SHAR, achieving accuracies 96.13% 87.85%, respectively. These results surpass those reported in state-of-the-art demonstrate effectiveness proposed activities.

Language: Английский

Citations

0