Mingyi Hong

Mingyi Hong is an Associate Professor in the Department of Electrical and Computer Engineering at the University of Minnesota, where he leads the OptimAI-Lab. His work bridges optimization theory, machine learning, and signal processing, with a focus on foundation models like LLMs and diffusion models.

• Education: Not explicitly mentioned in the text
• Current Projects: NSF grants on bilevel optimization, LLM unlearning, and inverse reinforcement learning

Research Themes:

• Bilevel Optimization: Applications in LLM alignment, unlearning, and wireless systems
• LLM Safety: Unlearning, alignment with human feedback, robustness
• Diffusion Models: Inference-time alignment, adversarial training
• Distributed Optimization: Privacy-preserving algorithms, federated learning

Recent Publications highlight trends in LLM unlearning (BLUR, LUME), optimization theory (Barrier Functions, νSAM), and diffusion models (Direct Noise Optimization). His group has secured NSF, AWS, Cisco, and Open Philanthropy grants.

Scientific Recognition:

• IEEE Fellow (2025)
• SPS Best Paper Award (2022, 2021, 2018)
• Doctoral Dissertation Fellowship (2024)
• IBM Pat Goldberg Memorial Award (2022)

He mentors PhD students like Siliang Zeng and Xinwei Zhang, and collaborates with institutions including Michigan State University, Amazon, and NIH on projects spanning UHF MRI technology to climate-smart agriculture.