James Beetham

About Me

I’m a fifth-year Ph.D. student in the Center for Research in Computer Vision (CRCV) at the University of Central Florida (UCF), working under Dr. Mubarak Shah. My research primarily focuses on adversarial attacks, with early work exploring unwilling knowledge distillation (model stealing) and more recent efforts centered on jailbreaking large language models (LLMs). Currently, I’m investigating how these techniques can be extended to the vision-language model (VLM) domain. I’m always open to discussing research or collaborating, feel free to reach out!

Research Interests

• Model Saftey: Exploring adversarial attacks on vision models and jailbreak techniques for large language models (LLMs), with a focus on both offensive (attack) and defensive (mitigation and red-teaming) strategies.

News

• [Apr. 2023] Presented model stealing paper @ ICLR 2023.
• [Jan. 2023] ICLR 2023 paper accepted for poster presentation.
• [Aug. 2022] Presented black-box white-box adversarial attack detection paper in Montreal @ ICPR 2022.
• [Mar. 2022] ICPR 2022 paper accepted for poster presentation.
• [Aug. 2020] Joined UCF CRCV lab.

Publications

Services

Paper Reviewing

• Neural Information Processing Systems 2025
• International Conference on Machine Learning
• Neural Information Processing Systems 2024

Mentoring for Research Experience for Undergrads (REU)

• Bypass data-free model stealing defenses 2024
• Detect and defend against model-stealing attacks 2022
• GAN guidance for salient adversarial attacks 2021

Powered by Jekyll and Minimal Light theme.