I am currently a Ph.D. student in Computer Science, specializing in artificial intelligence (AI) testing technology. After obtaining my Bachelor’s degree in Information and Computational Science from the School of Mathematics and Statistics at Beijing Institute of Technology in June 2019, I began my Master’s studies in Computer Software and Theory at the Academy of Military Sciences in September 2019. I transitioned to a doctoral program in April 2021 in the same field. My research primarily focuses on:

The safety testing of autonomous driving systems
Adversarial example research

These efforts aim to enhance the reliability and security of AI applications in critical areas.

🔥 News

2024.11：🎉 “Enhancing Adversarial Robustness through Self-Supervised Confidence-Based Denoising” has been accepted at the the 23rd IEEE TrustCom (TrustCom 2024, 110/555=19.82%).
2024.07：🎉 “Dance of the ADS: Orchestrating Failures through Historically-Informed Scenario Fuzzing” has been accepted at the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA 2024, 143/694=20.6%)!
2023.11: 🎉 “ADV-POST: Physically Realistic Adversarial Poster for Attacking Semantic Segmentation Models in Autonomous Driving” presented at the International Conference on Neural Information Processing (ICONIP 2023).
2023.09: 🎉 “Test Suite Generation Based on Context-Adapted Structural Coverage for Testing DNN” presented at the 2023 24th Asia-Pacific Network Operations and Management Symposium (APNOMS 2023), and received the “Student Best Paper Awards”!
2023.07: 🎉 “Driving into Danger: Adversarial Patch Attack on End-to-End Autonomous Driving Systems Using Deep Learning” presented at the 2023 IEEE Symposium on Computers and Communications (ISCC 2023).
2023.06: 🎉 “物理对抗补丁攻击与防御技术研究综述” accepted at Journal of Cyber Security (信息安全学报).
2023.06: 🎉 “Risk Scenario Generation for Autonomous Driving Systems based on Scenario Evaluation Model” presented at the 2023 International Joint Conference on Neural Networks (IJCNN 2023).

📝 Publications

🧑‍🎨 Selected Publications

IJCNN 2023

Risk Scenario Generation for Autonomous Driving Systems based on Scenario Evaluation Model

Tong Wang, Xiaohui Kuang, Huan Deng, Taotao Gu, Wei Kong, Jianwen Tian, Gang Zhao

Introducing a scenario evaluation model to enhance risk scenario generation.
Establishing a scenario-based testing framework using the CARLA Leaderboard to explore risk scenarios in autonomous driving systems.
Conducting a cross-sectional comparison of three open-source systems, LAV, Transfuser, and NEAT, to highlight safety distinctions.

@INPROCEEDINGS{10191164,
  author={Wang, Tong and Kuang, Xiaohui and Deng, Huan and Gu, Taotao and Kong, Wei and Tian, Jianwen and Zhao, Gang},
  booktitle={2023 International Joint Conference on Neural Networks (IJCNN)}, 
  title={Risk Scenario Generation for Autonomous Driving Systems based on Scenario Evaluation Model}, 
  year={2023},
  volume={},
  number={},
  pages={1-8},
  keywords={Deep learning;Neural networks;Closed box;Search problems;Robustness;Entropy;Safety},
  doi={10.1109/IJCNN54540.2023.10191164}}

ISCC 2023

Driving into Danger: Adversarial Patch Attack on End-to-End Autonomous Driving Systems Using Deep Learning

Tong Wang, Xiaohui Kuang, Hu Li, Qianjin Du, Zhanhao Hu, Huan Deng, Gang Zhao

Conducting a study on ADS using the CARLA Leaderboard, focusing on adversarial example generation.
Proposing a data collection theory for generating adversarial examples with post-realism optimization.
Developing an effective loss function targeting ADS through route planning and control signal information.
Implementing adversarial patches, specifically in the form of a T-shirt, to demonstrate dangerous behaviors in ADS under simulation.

@INPROCEEDINGS{10218291,
  author={Wang, Tong and Kuang, Xiaohui and Li, Hu and Du, Qianjin and Hu, Zhanhao and Deng, Huan and Zhao, Gang},
  booktitle={2023 IEEE Symposium on Computers and Communications (ISCC)}, 
  title={Driving into Danger: Adversarial Patch Attack on End-to-End Autonomous Driving Systems Using Deep Learning}, 
  year={2023},
  volume={},
  number={},
  pages={995-1000},
  keywords={Deep learning;Measurement;Pedestrians;Lighting;Behavioral sciences;Safety;Security;Autonomous driving systems;Deep learning techniques;Adversarial examples;Security vulnerabilities},
  doi={10.1109/ISCC58397.2023.10218291}}

ISSTA 2024

Dance of the ADS: Orchestrating Failures through Historically-Informed Scenario Fuzzing

Tong Wang, Taotao Gu, Huan Deng, Hu Li, Xiaohui Kuang, Gang Zhao

Unveiling ScenarioFuzz, a pioneering scenario-based fuzz testing methodology for autonomous driving systems.
Extracting essential data from map road networks to form a foundational scenario seed corpus for fuzz testing.
Integrating specialized mutators and mutation techniques with a graph neural network model to optimize the fuzzing process.
Reducing time cost by an average of 60.3% and increasing error scenarios discovered per unit of time by 103%.
Proposing a self-supervised collision trajectory clustering method to identify and summarize high-risk scenario categories.
Successfully uncovering 58 bugs across six tested systems, highlighting critical safety concerns of ADS.

@inproceedings{10.1145/3650212.3680344,
author = {Wang, Tong and Gu, Taotao and Deng, Huan and Li, Hu and Kuang, Xiaohui and Zhao, Gang},
title = {Dance of the ADS: Orchestrating Failures through Historically-Informed Scenario Fuzzing},
year = {2024},
isbn = {9798400706127},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3650212.3680344},
doi = {10.1145/3650212.3680344},
booktitle = {Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis},
pages = {1086–1098},
numpages = {13},
keywords = {Autonomous Driving Systems, Fuzzing, Scenario-based Testing},
location = {Vienna, Austria},
series = {ISSTA 2024}
}

📚 All Publications

📅 2024

Trustcom 2024 Enhancing Adversarial Robustness through Self-Supervised Confidence-Based Denoising, Yongkang Chen, Tong Wang, Wei Kong, Taotao Gu, Guiling Cao, Xiaohui Kuang. (CCF-C)
ISSTA 2024 Dance of the ADS: Orchestrating Failures through Historically-Informed Scenario Fuzzing, Tong Wang, Taotao Gu, Huan Deng, Hu Li, Xiaohui Kuang, Gang Zhao. (CCF-A)

📅 2023

ICONIP 2023 ADV-POST: Physically Realistic Adversarial Poster for Attacking Semantic Segmentation Models in Autonomous Driving, Huan Deng, Minhuan Huang, Tong Wang, Hu Li, Jianwen Tian, Qian Yan, Xiaohui Kuang. (CCF-C)
APNOMS 2023 Test Suite Generation Based on Context-Adapted Structural Coverage for Testing DNN, Wei Kong, Qianjin Du, Huayang Cao, Hu Li, Tong Wang, Jianwen Tian, Xiaohui Kuang. (CCF-C)
ISCC 2023 Driving into Danger: Adversarial Patch Attack on End-to-End Autonomous Driving Systems Using Deep Learning, Tong Wang, Xiaohui Kuang, Hu Li, Qianjin Du, Zhanhao Hu, Huan Deng, Gang Zhao. (CCF-C)
AdvML 2023 Toward Testing Deep Learning Library via Model Fuzzing, Wei Kong, Huayang Cao, Tong Wang, Yuanping Nie, Hu Li, Xiaohui Kuang.
信息安全学报 物理对抗补丁攻击与防御技术研究综述, 邓欢, 黄敏桓, 李虎, 王彤, 况晓辉. (CCF-B)
IJCNN 2023 Risk Scenario Generation for Autonomous Driving Systems based on Scenario Evaluation Model, Tong Wang, Xiaohui Kuang, Huan Deng, Taotao Gu, Wei Kong, Jianwen Tian, Gang Zhao. (CCF-C)

📅 2022

ICSIP 2022 DP2Net: Defense Against Adversarial Examples by Detecting Perturbed Pixels, Tong Wang, Haitian Liu, Feng Wang, Junchao Du, Yang Bao.

📅 2021

IJCNN 2021 An Efficient Method to Measure Robustness of ReLU-Based Classifiers via Search Space Pruning, Xinping Wang, Liangyu Chen, Tong Wang, Mingang Chen, Min Zhang. (CCF-C)
AEMCSE 2021 Safety Testing of Neural Network for Image Classification, Tong Wang, Feng Wang, YunZhi Xue, Yang Bao.

📅 2020

DSA 2020 A Coverage-Based Effective Data Augmentation Framework, Shuting Kang, Tong Wang, PengQi Wang, Feng Wang, Qian Dong, Guang Yang, LingZhong Meng, YunZhi Xue.

📖 Educations

2019.09 - Present, Ph.D., Academy of Military Sciences, China.
2015.09 - 2019.06, Bachelor, Beijing Institute of Technology, China.