A5100455413- Adversarial Robustness in Machine Learning
- Artificial Intelligence in Games
- Anomaly Detection Techniques and Applications
- Video Analysis and Summarization
- Digital Games and Media
- Reinforcement Learning in Robotics
- Advanced Neural Network Applications
- Privacy-Preserving Technologies in Data
- Advanced Malware Detection Techniques
- Sports Analytics and Performance
- Educational Games and Gamification
- Monetary Policy and Economic Impact
- Facility Location and Emergency Management
- Domain Adaptation and Few-Shot Learning
- Physical Unclonable Functions (PUFs) and Hardware Security
- Network Security and Intrusion Detection
- Global Financial Crisis and Policies
- Digital Media Forensic Detection
- Wireless Signal Modulation Classification
- Internet Traffic Analysis and Secure E-voting
- Neural Networks and Applications
- Complex Network Analysis Techniques
- Multi-Criteria Decision Making
- E-commerce and Technology Innovations
- Market Dynamics and Volatility
Beijing Institute of Technology
2010-2025
Shenyang Aerospace University
2014-2024
Ministry of Industry and Information Technology
2023-2024
Harbin Engineering University
2023-2024
Dalian University
2024
Dalian University of Technology
2017-2024
Westlake University
2024
Shantou University
2023
Guiyang Medical University
2023
Beijing Information Science & Technology University
2023
Adversarial transferability enables black-box attacks on unknown victim deep neural networks (DNNs), rendering viable in real-world scenarios. Current transferable create adversarial perturbation over the entire image, resulting excessive noise that overfit source model. Concentrating to dominant image regions are model-agnostic is crucial improving efficacy. However, limiting local spatial domain proves inadequate augmenting transferability. To this end, we propose a attack with...
Deep neural network (DNN) is applied widely in many applications and achieves state-of-the-art performance. However, DNN lacks transparency interpretability for users structure. Attackers can use this feature to embed trojan horses the structure, such as inserting a backdoor into DNN, so that learn both normal main task additional malicious tasks at same time. Besides, relies on data set training. tamper with training interfere process, attaching trigger input data. Because of defects...
This study proposed a rapid approach to determine the fuzzy number evaluate roof fall risk in coal mine. In assessments by theory, determination of numbers using triangular (TFN) incorporated into analytic hierarchy process is difficult and time-consuming task. A novel TFN was proposed. To reduce occurrence accidents, it necessary identify main factors assessments. nine-score table-type questionnaire adopted collected expert judgement, based on which language scales were determined, ranging...
Deep neural networks (DNNs) are increasingly used as the critical component of applications, bringing high computational costs. Many practitioners host their models on third-party platforms. This practice exposes DNNs to risks: A third party hosting model may use a malicious deep learning framework implement backdoor attack. Our goal is develop realistic potential for attacks in We introduce threatening and realistically implementable attack that highly stealthy flexible. inject trojans by...
Federated learning is a distributed machine approach that enables multiple participants to collaboratively train model without sharing their data, thus preserving privacy. However, the decentralized nature of federated also makes it susceptible backdoor attacks, where malicious can embed hidden vulnerabilities within model. Addressing these threats efficiently and effectively crucial, especially given impracticality iterative resource-intensive detection methods in environments. This article...
Machine learning has made tremendous progress and applied to various critical practical applications. However, recent studies have shown that machine models are vulnerable malicious attackers, such as neural network backdoor triggering. A successful triggering behavior may cause serious consequences, allowing the attacker bypass identity verification directly enter system. In image classification, there is always only one target label triggered by trigger in previous works. The position of...
Abstract Using microbiomes to mitigate global plastic pollution is of paramount importance. Insect have garnered emerging interest for their ability biodegrade non-hydrolysable polymers. The larvae Spodoptera frugiperda , a globally prevalent migratory crop pest, are accidentally discovered consume polyvinyl chloride (PVC) films, highlighting the role gut microbiome. Following migration S. frugiperd in China, this study displays comprehensive geographical profile its larval microbiota and...