Jinhao Liang

Hi, everyone. Welcome to my homepage.

I am currently a Ph.D. student in Computer Science at the University of Virginia (UVA), advised by Professor Ferdinando Fioretto. Before that, I completed my M.Phil degree in Computer and Information Engineering at The Chinese University of Hong Kong, Shenzhen (CUHK Shenzhen), under the supervision of Professor Chenye Wu. I also obtained a bachelor’s degree in Software Engineering at Xidian University (XDU).

Research

I am particularly interested in integrating generative AI with optimization to address complex scientific and engineering challenges. Currently, my research topics focus on developing algorithms to ensure the output of diffusion models/flow matching satisfying constraints with provably guarantees.

news

Nov 08, 2025	Paper titled “Discrete-Guided Diffusion for Scalable and Safe Multi-Robot Motion Planning” accepted to the Fortieth AAAI Conference on Artificial Intelligence (AAAI-26).
Sep 22, 2025	Our paper “Chance-constrained Flow Matching for High-Fidelity Constraint-aware Generation” was accepted at NeurIPS 2025 ML×OR Workshop and NeurIPS 2025 COML Workshop.
May 10, 2025	Our paper “Neuro-symbolic Generative Diffusion Models for Physically Grounded, Robust, and Safe Generation” was awarded a DARPA Disruptive Idea Award (Top 5 Conference Submission).
May 01, 2025	Our paper “Simultaneous Multi-Robot Motion Planning with Projected Diffusion Models” has been accepted by ICML 2025.
Apr 20, 2025	Our paper “Neuro-symbolic Generative Diffusion Models for Physically Grounded, Robust, and Safe Generation” accepted to NeuS 2025 and selected for Oral Presentation!
Dec 12, 2024	Paper titled “Multi-Agent Path Finding in Continuous Spaces with Projected Diffusion Models” accepted to AAAI-25 WoMAPF and AAAI-25 Bridge: AI+ORMS.
Nov 19, 2024	Honored to receive the CUHKSZ Presidential Award for Outstanding Graduate Students.
Feb 07, 2024	Honored to receive the UVA Provost Fellowship.
Jan 09, 2024	Our paper “Joint Chance-constrained Unit Commitment: Statistically Feasible Robust Optimization with Learning-to-Optimize Acceleration” has been accepted by IEEE Transactions on Power Systems.