Xiaoyu Li

Prof. Xiaoyu Li

orcid

Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 February 2022 in Journal of Colloid and Interface Science
Verified
Journal Article
Closed access
JGJia-Xing Guo
Jia-Xing Guo
SWShao-Yi Wu
Shao-Yi Wu
GZGao-Jun Zhang
Gao-Jun Zhang
LYLi Yan
Li Yan
JHJian-Guo Hu
Jian-Guo Hu
Xiaoyu LiXiaoyu Li
Xiaoyu Li

8 January 2025 in IEEE Transactions on Information Forensics and Security
Verified
Journal Article
Closed access
DZDesheng Zheng
Desheng Zheng
WKWuping Ke
Wuping Ke
Xiaoyu LiXiaoyu Li
Xiaoyu Li
YDYaoxin Duan
Yaoxin Duan
GYGuangqiang Yin
Guangqiang Yin
FMfan min
fan min
Adversarial examples have posed a serious threat to deep neural networks due to their transferability. Existing transfer-based attacks tend to improve the transferability of adversarial examples by destroying intrinsic features. However, prior work typically employed single-dimensional or additive importance estimates, which provide inaccurate representations of features. In this work, we propose the Multi-Feature Attention Attack (MFAA), which fuses multiple layers of feature representations to disrupt category-related features and thus improve the transferability of the adversarial examples. First, MFAA introduces a layer-aggregation gradient (LAG) to obtain guidance maps, which reflect the importance of features in multiple scales. Second, it generates ensemble attention (EA), preserving object-specific features and offsetting model-specific features based on the guidance maps. Third, EA is iteratively disturbed to achieve high transferability of the adversarial examples. Empirical evaluation on the standard ImageNet dataset shows that adversarial examples crafted by MFAA can effectively attack different networks. Compared to the state-of-the-art transferable attacks, our attack improves the average attack success rate of the black-box model with defense from 88.5% to 94.1% on single-model attacks and from 86.6% to 95.1% on ensemble attacks. Our code is available at Github: https://github.com/KWPCCC/MFAA.

8 May 2024 in Knowledge-Based Systems
Verified
Journal Article
Closed access
WKWuping Ke
Wuping Ke
DZDesheng Zheng
Desheng Zheng
Xiaoyu LiXiaoyu Li
Xiaoyu Li
YHYuanhang He
Yuanhang He
TLTianyu Li
Tianyu Li
FMfan min
fan min

21 October 2022 in Sensors
Verified
Journal Article
Open access
PDF file
Collins SeyCollins Sey
Collins Sey
HLHang Lei
Hang Lei
Weizhong QianWeizhong Qian
Weizhong Qian
Xiaoyu LiXiaoyu Li
Xiaoyu Li
LFLinda Delali Fiasam
Linda Delali Fiasam
IAIsaac Osei Agyemang
Isaac Osei Agyemang
The rapid advancement of the Internet of Vehicles (IoV) has led to a massive growth in data received from IoV networks. The cloud storage has been a timely service that provides a vast range of data storage for IoV networks. However, existing data storage and access models used to manage and protect data in IoV networks have proven to be insufficient. They are centralized and usually accompanied by a lack of trust, transparency, security, immutability, and provenance. In this paper, we propose VBlock, a blockchain-based system that addresses the issues of illegal modification of outsourced vehicular data for smart city management and improvement. We introduce a novel collusion-resistant model for outsourcing data to cloud storage that ensures the network remains tamper-proof, has good data provenance and auditing, and solves the centralized problems prone to the single point of failure. We introduced a key revocation mechanism to secure the network from malicious nodes. We formally define the system model of VBlock in the setting of a consortium blockchain. Our simulation results and security analysis show that the proposed model provides a strong security guarantee with high efficiency and is practicable in the IoV environment.

22 August 2024 in Expert Systems with Applications
Verified
Journal Article
Closed access
LYLian-Hui Yu
Lian-Hui Yu
Xiaoyu LiXiaoyu Li
Xiaoyu Li
GCGeng Chen
Geng Chen
QZQin-Sheng Zhu
Qin-Sheng Zhu
HLHui Li
Hui Li
GYGuowu Yang
Guowu Yang

1 January 2020 in Computers, Materials & Continua
Verified
Journal Article
Open access
YHYi-Ming Huang
Yi-Ming Huang
HLHang Lei
Hang Lei
Xiaoyu LiXiaoyu Li
Xiaoyu Li
QZQing-Sheng Zhu
Qing-Sheng Zhu
WRWang-Hao Ren
Wang-Hao Ren
XLXu-Sheng Liu
Xu-Sheng Liu
In recent years, an increasing number of studies about quantum machine learning not only provide powerful tools for quantum chemistry and quantum physics but also improve the classical learning algorithm. The hybrid quantum-classical framework, which is constructed by a variational quantum circuit (VQC) and an optimizer, plays a key role in the latest quantum machine learning studies. Nevertheless, in these hybridframework-based quantum machine learning models, the VQC is mainly constructed with a fixed structure and this structure causes inflexibility problems. There are also few studies focused on comparing the performance of quantum generative models with different loss functions. In this study, we address the inflexibility problem by adopting the variable-depth VQC model to automatically change the structure of the quantum circuit according to the qBAS score. The basic idea behind the variable-depth VQC is to consider the depth of the quantum circuit as a parameter during the training. Meanwhile, we compared the performance of the variable-depth VQC model based on four widely used statistical distances set as the loss functions, including Kullback-Leibler divergence (KL-divergence), Jensen-Shannon divergence (JS-divergence), total variation distance, and maximum mean discrepancy. Our numerical experiment shows a promising result that the variable-depth VQC model works better than the original VQC in the generative learning tasks.