Xin Wang
Ph.D. Student, Transportation Engineering — University of Washington
About
My research focuses on making AI safer and more reliable, enabling trustworthy AI deployment in safety-critical transportation applications, for example:
- AI Stability — How can small changes in data or settings affect the outcomes of deep learning models, and how can we make them more predictable and trustworthy?
- Machine Unlearning — How can we allow AI systems to "forget" specific data when needed, helping protect privacy and meet new regulations without retraining from scratch?
- Adversarial Robustness — How can malicious or faulty data trick AI, and how can we design defenses that make models resistant to these attacks?
- Applications in Transportation — How can these AI techniques be applied to traffic forecasting, connected vehicles, and intelligent transportation systems to help cities manage mobility more safely and efficiently?
AI Stability
Machine Unlearning
Adversarial Robustness
Trustworthy Transportation
Research Focus
AI
- Set-to-set analysis & Lipschitz-like stability
- Machine unlearning on constrained learning
- Variational inequalities, influence functions, sensitivity analysis
- PINNs and multi-physics modeling
Transportation
- Adversarial data poisoning on traffic forecasting
- Robust defenses for signal control / ramp metering
- Unlearning for trajectory data and incident response
- Trustworthy AI for CAVs and V2X communications
Education
- Ph.D., Transportation Engineering, University of Washington, 2022—present — Advisor: Prof. Xuegang (Jeff) Ban
- M.S., Statistics, Renmin University of China, 2020–2022
- B.S., Applied Mathematics, Central South University, 2016–2020
Experience
- Research Assistant, University of Washington, 2022—present
- Teaching Assistant, CET 513: Optimization in Transportation (Autumn 2025)
- Teaching Assistant, CET 513: Optimization in Transportation (Autumn 2024)
- ML Engineer Intern, Baidu Inc., Jan–May 2021 — Multi-objective ranking (PE-LTR, NSGA-II) improving NDCG and CTR
Academic Service
- Reviewer: Transportation Research Part C, TRB Annual Meeting, AAAI
Publications
- AI
-
Set-Valued Sensitivity Analysis of Deep Neural Networks
Paper
Xin Wang, Feilong Wang, Xuegang (Jeff) Ban
AAAI 2025, 39(20): 21304–21311 -
Machine Unlearning of Traffic State Estimation and Prediction
Paper
Xin Wang, R. Tyrrell Rockafellar, Xuegang (Jeff) Ban
arXiv:2507.17984 (2025). -
Model-Targeted Data Poisoning Attacks against ITS Applications with Provable Convergence
Paper
Xin Wang, Feilong Wang, Yuan Hong, R. Tyrrell Rockafellar, Xuegang (Jeff) Ban
arXiv:2505.03966 (2025). - Transportation
-
Data poisoning attacks on traffic state estimation and prediction
Paper
Feilong Wang, Xin Wang, Yuan Hong, R. Tyrrell Rockafellar, Xuegang (Jeff) Ban
Transportation Research Part C, 168 (2024): 104577 -
Data poisoning attacks in intelligent transportation systems: A survey
Paper
Feilong Wang, Xin Wang, Xuegang (Jeff) Ban
Transportation Research Part C, 165 (2024): 104750 -
Infrastructure-enabled Defense Methods against Data Poisoning Attacks on Traffic State Estimation and Prediction
Paper
Feilong Wang, Xin Wang, Yuan Hong, Xuegang (Jeff) Ban
TRC-30, 2025 -
Transferability in Data Poisoning Attacks on Spatiotemporal Traffic Forecasting Models
Paper
Xin Wang, Feilong Wang, Yuan Hong, Xuegang (Jeff) Ban
Transportation Research Part C, 183 (2026): 105501
Invited Talks & Guest Lectures
- AAAI 2025 (Poster) — Set-Valued Sensitivity Analysis of Deep Neural Networks
- ISTTT25 — Data Poisoning Attacks on Traffic State Estimation and Prediction
- TRB 2025 — A Review of Data Poisoning Attacks in Intelligent Transportation Systems