I’m a PhD candidate specializing in AI/ML and I like building new things.

I build end-to-end machine learning pipelines—from model training to deployment and monitoring—that solve high-impact safety and infrastructure problems, with a focus on real-time prediction, anomaly detection, and reinforcement learning systems.

I built Cyber-TFWS (a real-time cyberattack forecasting & warning system for connected vehicles) and RL Signal Control (a deep RL traffic signal agent that improves safety and congestion using simulation). I enjoy open-source, applied research, and systems that make experiments reproducible, scalable, and fast.

Experience

Since [Date], I have been a [position] at [Company/Organization], [description]. My contribution includes:

Major project 1 – description
Major project 2 – description
Major project 3 – description
Major project 4 – description
Major project 5 – description
Trajectory-Based Cyberattack Detection for Connected Vehicles – Built a machine learning system using Hidden Markov Models (HMM) to detect spoofing cyberattacks at connected signalized intersections based on vehicle trajectory patterns. Achieved 98% detection accuracy and demonstrated superior performance compared with CNN and Bi-LSTM models. The system analyzes velocity, acceleration, and distance-to-stop-line data to identify abnormal driving behaviors caused by falsified signal messages.
Human-in-the-Loop Cyberattack Simulation Platform – Designed a cyberattack simulation environment using a driving simulator with 32 participants to collect real human driving trajectories under normal and attack scenarios. Developed a dataset of 96 trajectory samples capturing driver reactions to falsified red-light countdown messages, enabling research on human behavior and anomaly detection in connected vehicle environments.
HMM-4-C Cyberattack Detection Algorithm – Developed a probabilistic anomaly detection model based on Hidden Markov Models and Expectation–Maximization (EM) training to identify cyberattack states in vehicle trajectory time series. The model uses transition and emission probabilities to classify three states: Pass, Stop, and Cyberattack, providing robust detection with limited training data.
Trajectory-Based Real-Time Cyberattack Warning System – Proposed a roadside sensor detection architecture capable of capturing vehicle trajectories and performing real-time cyberattack detection without relying on vulnerable communication channels. The system computes attack probability and triggers warning messages to nearby vehicles and traffic authorities when anomalies are detected.
Early Cyberattack Detection Using Effective Detection Period (EDP) and Distance (EDD) – Introduced two novel metrics, Effective Detection Period (EDP) and Effective Detection Distance (EDD), to determine how quickly a cyberattack can be detected using vehicle trajectory data. The analysis shows cyberattacks can be detected within 3–6 seconds after the anomaly occurs and within 50 ft of the intersection, enabling faster safety interventions.

For the [period], I was a [position] at Organization, where I worked with [colleagues] on [project description].

For the [period], I was a [position] at Organization, where I worked on [project description].

Education

Ph.D. in Transportation Engineering, University of Cincinnati
M.S. in Computer Science, Georgia Institute of Technology
B.S./M.S. in Engineering, Nanjing Uniersity

Publications

Paper Title [Authors] ([Year]) [Journal/Conference Name]

Another Paper Title [Authors] ([Year]) [Journal/Conference Name]

Major Publications

A Cyberattack Warning System for Enhancing Connected Vehicle Safety under Spoofing Cyberattacks
Gu, Y., Li, Z., Wei, H., Zhang, G., Xu, Y. (2025)
IEEE Transactions on Intelligent Transportation Systems (IF: 8.4)
→ Proposed a real-time GAN-based human-in-the-loop prediction system that improves early cyberattack detection and enhances intersection safety.

Why to Buy or Why Not to Buy? Automated Vehicle Purchase Behavior Analysis
Gu, Y., Wang, S., Li, Z., Zhang, G., Ai, C., Li, P. (2025)
Research in Transportation Economics (IF: 4.6)
→ Built structural equation models to uncover psychological and behavioral drivers behind automated vehicle adoption.

Public Preferences and Concerns Regarding Automated Vehicle-Based Transportation Services: A Mechanism Analysis from a Kentucky Survey
Wang, S., Li, Z., Wang, Y., Zhao, W., Gu, Y., & Wei, H. (2025)
Transportation Letters, 17(3), 502–511

Oral Presentations

Cybersecure Transit: Innovation at the Intersection of AI, IoT, and Smart Mobility
Gu, Y. (2024)
Presented an AI-driven system for detecting cyberattacks at interconnected signalized intersections.
USDOT Tier 1 UTC Transportation Cybersecurity Center for Advanced Research and Education (CYBER-CARE)
The Morrison-Clark Historic Inn & Restaurant, Washington, DC
January 10, 2024

Event Website: CYBER-CARE event page

Conference

Safety Warning System for Connected Vehicles under Spoofing Cyberattacks at a Connected Signalized Intersection
Gu, Y., Li, Z., Wei, H., Zhang, G., Xu, Y.
The 104th Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 2025

Modeling Dynamic Vehicle-Driver Complex Behaviors at Signalized Intersections Under Cyberattacks
Xu, Y., Li, Z., Wei, H., Zhang, G., Gu, Y.
The 104th Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 2025

Smart Leading Pedestrian Intervals (SLPIs): A Deep Reinforcement Learning Control Strategy for Determining Optimal LPIs
Gu, Y., Li, Z.
The 103rd Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 2024

Machine Learning-Driven Dynamic Trajectory Planning to Support Human-like Automated Driving for Stop-Controlled Local Roads
Nian, D., Li, Z., Gu, Y., Kluger, R., Wei, H., Lin, W.
The 103rd Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 2024

Advanced Radar Sensing-Based Investigation of Stop Sign Spacing’s Impact on Vehicle Speeds through Speed Trajectory Analysis
Nian, D., Li, Z., Gu, Y., Kluger, R., Wei, H., Lin, W.
The 103rd Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 2024

Machine Learning-Based Detection of Cyberattacks at Connected Signalized Intersections Using Hidden Markov Models
Gu, Y., Li, Z., Zhang, Y., Tiwari, S., Wang, S.
The 102nd Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 2023

Revealing the Hidden Factors that Influence People’s Willingness to Purchase Automated Vehicles: A Mediation Analysis Based on Structural Equation Modeling
Gu, Y., Li, Z.
The 102nd Annual Meeting of the Transportation Research Board (TRB), Washington, D.C., 2023

Enhancing Connected Vehicle Safety via a Trajectory-Forecasting-Based Cyberattack Warning System
Gu, Y.
CYBER-CARE Annual Symposium 2024: Safeguarding Transportation Cybersecurity in the Digital Age
Houston, Texas, 2024
Program: https://www.uh.edu/cybercare/_files/cyberprogram.pdf

Miscellaneous

Photography / Computer Graphics / Web Development