|
Ruofeng Yang (杨若峰)
|
About me
I am a 4-year PhD student in computer science in John Hopcroft Center, Shanghai Jiao Tong University under the supervision of Prof. Shuai Li since Sept. 2022.
My research focuses on diffusion models and reinforcement learning. Previously, I worked on the theoretical analysis of diffusion models and reinforcement learning algorithms. Currently, I am interested in integrating diffusion models with reinforcement learning, including reinforcement learning fine-tuning (RLFT) for 3D generation (IEG, Tencent Rhino-Bird Research Elite Program, 犀牛鸟精英人才计划) and video generation (美团北斗人才计划, Longcat-video base model team).
His research focuses on computer vision generation, particularly on the following topics:
- Diffusion Models (Image, Video, Theory, Post Training)
- Deep Learning Theory
- Reinforcement Learning and RLHF
He anticipates graduating in 2027 for industrial research positions! If you're interested, please feel free to reach out via email or WeChat (yrf13618645542).
Research Experiences
- Intern at the National University of Singapore (NUS), School of Computing, 2026/01-2026/07, Singapore.
- Intern at Meituan, Longcat-Video Team as a member of the 北斗人才计划.
- Intern at Tencent IEG, Game AI center as a member of the Tencent Rhino-Bird Research Elite Program (犀牛鸟精英人才计划), 2023, Shenzhen.
News
- I post a blog on the relationship between diffusion and representation learning and manifold learning [html][pdf][slides]
- One paper about the MoE structure of Diffusion Models is accepted by ICLR 2026!
- I provide a Openreview plugin for ICLR 2026!
- One paper about consistency models is accepted by ICML 2025!
- One paper about improved iteration complexity of VESDE (reverse SDE and PFODE) is accepted by AISTATS 2025!
- Two papers about the iteration complexity of diffusion models (including a training-free method) and few-shot diffusion models are accepted by NeurIPS 2024!
- One paper is accepted by NeurIPS 2023!
- One paper is accepted by ICLR 2023!
- One paper is accepted by AAAI 2023 (Oral)!
Blogs & Tutorials & Talks
- Blog on the relationship between diffusion and representation learning and manifold learning (Theory & Application)[html][pdf][slides]
- A simple introduction to diffusion model (Application)
- Why Rectified Flow is Better? Elucidating VP, VE, and RF-based diffusion models (Theory)
- A brief survey on some interesting directions about diffusion models (Theory) (Coming Soon)
Publications
As we know, diffusion models can be roughly divided into pretraining, supervised fine-tuning, RL posting training, and sampling algorithm design. My works focus on these four areas.
Diffusion Model MoE Structure and Pretraining
|
Multi-Subspace Multi-Modal Modeling for Diffusion Models: Estimation, Convergence and Mixture of Experts (ICLR 2026)
Ruofeng Yang, Yongcan Li (Equal Contribution), Bo Jiang, Cheng Chen, Shuai Li [ICLR 2026] |
Diffusion Model Supervised Fine-tuning
|
Few-shot Diffusion Models Escape the Curse of Dimensionality (NeurIPS 2024)
Ruofeng Yang, Bo Jiang, Cheng Chen, Ruinan Jin, Baoxiang Wang, Shuai Li [Best paper award (2nd Prize), TongAI 2025] [NeurIPS 2024] |
|
Evaluating the Role of Great Pre-trained Diffusion Models in Few-shot Phase: Warm-up and Acceleration
Ruofeng Yang, Yongcan Li (Equal Contribution) Bo Jiang, Cheng Chen, Shuai Li [Preprint] |
Diffusion Model Post-Training
|
Contrastive guidance and feedback: A Suitable way to improve 3D Consistency of Multi-view Diffusion Model
Ruofeng Yang, Le Wan, Yaqing Zhang, Shuai Li [Preprint] |
Diffusion Model Sampling and Condition Generation
|
Elucidating Rectified Flow with Deterministic Sampler: Polynomial Discretization Complexity for Multi and One-step Models
Ruofeng Yang, Zhaoyu Zhu, Bo Jiang, Cheng Chen, Shuai Li A talk on [CSML 2025] [Preprint] |
|
|
Improved Discretization Complexity Analysis of Consistency Models: Variance Exploding Forward Process and Decay Discretization Scheme (ICML 2025)
Ruofeng Yang, Bo Jiang, Cheng Chen, Shuai Li [ICML 2025] |
|
The Polynomial Iteration Complexity for Variance Exploding Diffusion Models: Elucidating SDE and ODE Samplers (AISTATS 2025)
Ruofeng Yang, Bo Jiang, Shuai Li [AISTATS 2025] |
|
Leveraging Drift to Improve Sample Complexity of Variance Exploding Diffusion Models (NeurIPS 2024)
Ruofeng Yang, Zhijie Wang, Bo Jiang, Shuai Li [NeurIPS 2024] |
|
Elucidating Guidance in Variance Exploding Diffusion Models: Fast Convergence and Better Diversity
Ruofeng Yang, Qiuyi Yu, Bo Jiang, Shuai Li [Preprint] |
Representation and Reinforcement Learning Theory
|
Understanding Representation Learnability of Nonlinear Self-Supervised Learning (AAAI 2023 Oral)
Ruofeng Yang, Xiangyuan Li, Bo Jiang, Shuai Li [Code] [AAAI 2023] |
|
Learning Adversarial Linear Mixture Markov Decision Processes with Bandit Feedback and Unknown Transition (ICLR 2023)
Canzhe Zhao, Ruofeng Yang, Baoxiang Wang, Shuai Li [ICLR 2023] |
|
Learning Adversarial Low-rank Markov Decision Processes with Unknown Transition and Full-information Feedback (NeurIPS 2023)
Canzhe Zhao, Ruofeng Yang, Baoxiang Wang, Xuezhou Zhang, Shuai Li [NeurIPS 2023] |
Professional Services
Conference Reviewer for
- International Conference on Machine Learning (ICML) 2025
- International Conference on Artificial Intelligence and Statistics (AISTATS) 2025
- International Conference on Learning Representations (ICLR) 2025
- Neural Information Processing Systems (NeurIPS) 2024
- Autonomous Agents and Multiagent Systems (AAMAS) 2022
Rewards
- National Scholarship (for Ph.D. students), from the Ministry of Education of China
- Shanghai Jiao Tong University Outstanding Graduates 2022
- Outstanding Winner of Mathematical Contest in Modeling (top 0.1%) 2020
- Tung Scholarship (Hong Kong's Tung Foundation) 2020
- Yang You Scholarship 2019, 2021