Machine Learning Ph.D. student @ Georgia Tech
Wei Guo郭纬
I study diffusion-based generative models and sampling algorithms, with a current focus on continuous diffusion LLMs.
About Me
I am a Machine Learning Ph.D. student at Georgia Institute of Technology starting from fall 2023, advised by Professors Yongxin Chen and Molei Tao. Previously, I obtained my Bachelor's degree in Statistics from the School of Mathematical Sciences, Peking University in 2023, where I was mentored by Professor Cheng Zhang. I was born in Ningbo, Zhejiang Province, P.R. China in 2001, and grew up there until I left for Beijing in 2019.
CV
For a complete overview of my education, publications, and research experience, please check my CV.
Academic Interests
I am broadly interested in fields ranging from statistics, probability, and machine learning. My current research interests include but are not limited to:
I think about fundamental ways of generating data. In particular, I focus on (continuous and discrete) diffusion/flow-based models with applications to vision, language, and scientific domains.
Due to my math background, I'm also interested in the theoretical analysis and practical design of sampling algorithms, including Markov chain Monte Carlo, non-equilibrium (e.g., denoising diffusion, stochastic localization) methods, and learning-based neural samplers. Meanwhile, I also work on applied stochastic analysis with connections to optimal transport, stochastic optimal control, and statistical physics.
News
Selected Papers
2026
-
ICLR 2026Complexity Analysis of Normalizing Constant Estimation: from Jarzynski Equality to Annealed Importance Sampling and beyondIn The Fourteenth International Conference on Learning Representations, 2026TL;DRGives a unified complexity analysis for normalizing constant estimation methods: from Jarzynski equality to annealed importance sampling and reverse diffusion sampler. -
PreprintContinuous Diffusion Scales Competitively with Discrete Diffusion for LanguagearXiv preprint arXiv:2605.18530, 2026TL;DRWe establish the first scaling law for continuous diffusion language models (DLMs) that rivals discrete DLMs. -
ICML 2026 SpotlightEnhancing Reasoning for Diffusion LLMs via Distribution Matching Policy OptimizationIn Forty-third International Conference on Machine Learning (Spotlight, top 2.2%), 2026TL;DRDistribution matching policy optimization via weighted denoising cross-entropy: a new RL paradigm beyond policy gradients.
2025
-
ICLR 2025Provable Benefit of Annealed Langevin Monte Carlo for Non-log-concave SamplingIn The Thirteenth International Conference on Learning Representations, 2025TL;DRProvides a theoretical guarantee of the convergence of annealed Langevin Monte Carlo from the perspective of optimal transport and Girsanov’s theorem. -
NeurIPS 2025MDNS: Masked Diffusion Neural Sampler via Stochastic Optimal ControlIn The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025TL;DRFrames masked diffusion neural sampling as a stochastic optimal control problem for efficient discrete neural sampler training.
Beyond Research
I am also interested in languages (including: la langue française (the French language, which I have been learning for one and a half year at PKU), vernaculars of Chinese and minority languages in China, and linguistics), political science, China's railway system, civil aviation, architecture, and video games (in particular, action games such as The Last of Us, Assassin's Creed, and the Метро (Metro) series). I am a fan of LE SSERAFIM. Finally, I am an enthusiast of traveling. Some of my highly recommended destinations that I have been to, and possessing profound historical and cultural heritage, include Hangzhou, Yangzhou, Datong, Macau, Washington, D.C., and my hometown Ningbo.