Zhengbang Yang
I am a Ph.D. student in Cybersecurity at George Mason University, advised by Dr. Zhuangdi Zhu. My research focuses on LLM alignment, enhancing the safety and accountability of large language models (LLMs) via reinforcement learning-based alignment and unlearning, and systematic design evaluation of agentic AI systems.
Before joining George Mason University, I earned my M.S. in Data Science from the University of California, Irvine, and dual bachelor’s degrees in Computer Science and Finance from Shenzhen University.
Education
- George Mason University, Ph.D. in Cybersecurity, 2024 - Present
- University of California, Irvine, M.S. in Data Science, 2022 - 2023
- Shenzhen University, B.Eng. in Computer Science and B.Econ. in Finance, 2018 - 2022
Research Interests
- Large Language Model (LLM) and Vision Language Model (VLM)
- Agentic AI for Health
- Security and Privacy
News
- Aug. 2026: 📃 First-authored paper CALIBURN: Self-Calibrated LLM Unlearning Alignment has been accepted to the EMNLP 2026 Main Conference.
- Aug. 2026: 📃 Our paper Distribution Corrected Offline Data Distillation for Large Language Models has been accepted to the Efficient Reasoning Workshop at COLM 2026.
- June 2026: 📃 First-authored paper Can LLMs Prove Robotic Path Planning Optimality? A Benchmark for Research-Level Algorithm Verification has been accepted to IEEE IROS 2026.
- Jan. 2026: 📃 Our paper DUET: Distilled LLM Unlearning from an Efficiently Contextualized Teacher has been accepted to ICLR 2026.
- Dec. 2025: 📃 First-authored paper CATNIP: LLM Unlearning via Calibrated and Tokenized Negative Preference Alignment has been accepted to the ResponsibleFM Workshop at NeurIPS 2025.
- May 2025: 📃 Our paper Hierarchical Federated Unlearning for Large Language Models has been accepted to FedKDD 2025.
- May 2025: 📃 First-authored paper ChatWise: A Strategy-Guided Chatbot for Enhancing Cognitive Support in Older Adults has been accepted to the KDD 2025 SciSocLLM Workshop.
- Jan. 2025: 📃 Our paper SPORTU: A Comprehensive Sports Understanding Benchmark for Multimodal Large Language Models has been accepted to ICLR 2025.
- Mar. 2024: 📃 Our paper SportQA: A Benchmark for Sports Understanding in Large Language Models has been accepted to NAACL 2024.