A5102951858- Complex Network Analysis Techniques
- Opportunistic and Delay-Tolerant Networks
- Fault Detection and Control Systems
- Opinion Dynamics and Social Influence
- Adversarial Robustness in Machine Learning
- Network Security and Intrusion Detection
- Network Time Synchronization Technologies
- Bacillus and Francisella bacterial research
- Anomaly Detection Techniques and Applications
- Advanced Computational Techniques and Applications
- Power Systems and Technologies
- Advanced Optical Network Technologies
- Hybrid Renewable Energy Systems
- Human Pose and Action Recognition
- Hand Gesture Recognition Systems
- Mobile Agent-Based Network Management
- Network Traffic and Congestion Control
- Graph theory and applications
- Smart Grid and Power Systems
- Mineral Processing and Grinding
- Peer-to-Peer Network Technologies
- Physical Unclonable Functions (PUFs) and Hardware Security
- Power Line Communications and Noise
- Advanced Decision-Making Techniques
- Topological and Geometric Data Analysis
State Key Laboratory of Industrial Control Technology
2021-2024
China Power Engineering Consulting Group (China)
2024
Beijing Normal University
2012-2023
Hohai University
2020-2023
State Key Joint Laboratory of Environment Simulation and Pollution Control
2021-2023
Zhejiang Sci-Tech University
2023
University of Waterloo
2023
China Southern Power Grid (China)
2021-2022
Zhejiang University of Technology
2021-2022
Peng Cheng Laboratory
2021
The label scarcity problem widely exists in industrial processes. In particular, samples of some fault types are extremely rare; even worse, the certain faults cannot be accessed, but they may appear actual process. These two kinds challenges together can termed as any-shot learning diagnosis. this article, taking advantages generative adversarial network, a approach is proposed to tackle problem, which generates abundant for those rare and inaccessible faults, trains strong diagnosis model....
In modern industries, data-driven fault detection and classification (FDC) systems can efficiently maintain industrial security stability, while the of FDC system itself is rarely or even never considered. The problem named adversarial vulnerability intrinsic machine learning models, which will give incorrect predictions under maliciously perturbed input data. This paper presents a work on this new topic systems, by 1) summarizing comparing various recent typical attack defense methods for...
Feature attribution explains Artificial Intelligence (AI) at the instance level by providing importance scores of input features' contributions to model prediction. Integrated Gradients (IG) is a prominent path method for deep neural networks, involving integration gradients along from explained (explicand) counterfactual (baseline). Current IG variants primarily focus on gradient explicand's output. However, our research indicates that output significantly affects feature as well. To...
Abstract In this study, we first inversely derive the required Willis parameters for realizing extreme contrast of scattering waves in opposite direction. Then, propose an active acoustic metastructure to achieve nonreciprocal coupling and experimentally realize strong transmission reflection simultaneously one direction while exhibiting absorption Without prior knowledge incident field, working frequency can be self-adjusted real time. The proposed provides avenue on unsurpassed control...
Soil pollution is a serious threat to human life and development. Different remedial measures are applied soils with different levels of contamination. The degree soil contamination in areas generally evaluated categorised based on the analysis samples. Regional sampling sites sparse because cost other factors, which makes it difficult accurately assess extent regional In this study, spatial classification model was established for Di(2-ethylhexyl) phthalate (DEHP) level using Bi-directional...
The routing strategy is important to information dissemination and exchange in complex communication networks. traditional strategies are often insufficient alleviate congestion enhance the utilization efficiency of two-layer In this paper, a novel efficient static weighted proposed. weight logical edges mapped on hub nodes at physical layer assigned higher according degree nodes. Then packets choose path with minimum sum edges. It found that average packet transmission time throughput have...
We investigate the robustness of complex communication networks on allocating redundancy links. The protecting key nodes (PKN) strategy is proposed to improve against intentional attack. Our numerical simulations show that a few redundant links among using PKN will significantly increase scale-free networks. have also theoretically proved and demonstrated effectiveness strategy. expect our work help achieve better understanding
An edge compensatory model with the tunable parameter α to guide process of enhancing robustness scale-free (SF) networks by adding new edges is presented. In addition, ability different resist intentional attacks, simulated as breakdown a group network nodes chosen according degree, explored. Simulations show that against attack can be improved when α< 2. Once decreased − 2, reducing further does little enhance SF networks. Moreover, novel information protection strategy proposed. Empirical...
Recently, data-driven industrial monitoring systems have been rapidly developed and significantly improved the performance on tasks. However, widely deployed models expose data to more unsecured links, which increase safety risk of safe-critical systems. The research about adversarial attacks has shown that potential attackers can utilize tiny crafted perturbations input mislead machine learning models' output. In this article, a novel cross-domain protection scheme named "Data Guardian" is...
Data-driven fault detection and classification (FDC) systems play an important role in ensuring the stability security of modern industry. However, issue data-driven FDC itself poses new challenges, where model prediction can be seriously damaged by maliciously manipulated imperceptible perturbations, known as adversarial attack. Since attacks may threaten models even whole safety-critical industrial systems, there is urgent need for guarantee security. In this article, a scheme presented...
To improve robustness for the complex network, this paper propose a novel probabilistic approach to allocate redundancy links based on nodal degree, i.e. preferential allocation, inverse and hiding (HRL) strategy. We find that allocating can efficiently of networks tolerate coordinated attack. Especially, I-I case, which both source node destination are chosen by performs best. Furthermore, we investigate effects HRL strategy under attack, simulation shows achieve better performance than...
This paper propose a novel probabilistic approach to allocate redundancy links and hiding (HRL) strategy improve robustness of complex communication networks. We find that allocating can efficiently against coordinated attack. The simulation shows HRL achieve the better performance.