Yunze Tong

Hi, I’m Yunze Tong, a third-year PhD student at Zhejiang University in Professor Fei Wu’s group.

My current research focuses on two directions:

Video Generation and Editing: Primarily centered on Streaming Video Models, with a growing interest in world models.
Reinforcement Learning for Generative Models: Leveraging RL to enhance denoising performance by optimizing both the denoising trajectory and the initial noise distribution.

In the earlier stages of my PhD, I also worked on model interpretability (specifically SHAP and its variants) and Out-of-Distribution (OOD) generalization, with a focus on covariate shift.

I earned my Bachelor’s degree from Zhejiang University in 2023 under the supervision of Professor Kun Kuang. During my studies, I also had the opportunity to work with Professor Jun Xiao.

I am always open to academic discussions—please feel free to reach out!

news

Jan 2026	One paper accepted at ICLR 2026. Grateful to all my co-authors for their hard work and support. 🎊
Oct 2025	Attending ICCV 2025 in Honolulu, Hawaii. 🌺

selected publications

Preprint

Noise Projection: Closing the Prompt-Agnostic Gap Behind Text-to-Image Misalignment in Diffusion Models

Yunze Tong, Didi Zhu, Zijing Hu, and 2 more authors

2025

Bib

@misc{tong2025noiseprojectionclosingpromptagnostic,
  title = {Noise Projection: Closing the Prompt-Agnostic Gap Behind Text-to-Image Misalignment in Diffusion Models},
  author = {Tong, Yunze and Zhu, Didi and Hu, Zijing and Yang, Jinluan and Zhao, Ziyu},
  year = {2025},
  eprint = {2510.14526},
  archiveprefix = {arXiv},
  primaryclass = {cs.CV},
}

ICLR

Asynchronous Denoising Diffusion Models for Aligning Text-to-Image Generation

Zijing Hu, Yunze Tong, Fengda Zhang, and 3 more authors

2025

Bib

@misc{hu2025asynchronousdenoisingdiffusionmodels,
  title = {Asynchronous Denoising Diffusion Models for Aligning Text-to-Image Generation},
  author = {Hu, Zijing and Tong, Yunze and Zhang, Fengda and Yuan, Junkun and Xiao, Jun and Kuang, Kun},
  year = {2025},
  eprint = {2510.04504},
  archiveprefix = {arXiv},
  primaryclass = {cs.CV},
}

ICCV

Decoding Correlation-Induced Misalignment in the Stable Diffusion Workflow for Text-to-Image Generation

Yunze Tong, Fengda Zhang, Didi Zhu, and 2 more authors

In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2025

Bib Code

@inproceedings{tong2025decoding,
  title = {Decoding Correlation-Induced Misalignment in the Stable Diffusion Workflow for Text-to-Image Generation},
  author = {Tong, Yunze and Zhang, Fengda and Zhu, Didi and Xiao, Jun and Kuang, Kun},
  booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
  year = {2025},
}

ICML

Latent Score-Based Reweighting for Robust Classification on Imbalanced Tabular Data

Yunze Tong, Fengda Zhang, Zihao Tang, and 5 more authors

In Proceedings of the 42nd International Conference on Machine Learning, 2025

Bib Code

@inproceedings{tong25latent,
  title = {Latent Score-Based Reweighting for Robust Classification on Imbalanced Tabular Data},
  author = {Tong, Yunze and Zhang, Fengda and Tang, Zihao and Gao, Kaifeng and Huang, Kai and Lyu, Pengfei and Xiao, Jun and Kuang, Kun},
  booktitle = {Proceedings of the 42nd International Conference on Machine Learning},
  year = {2025},
}

education

Ph.D. in Computer Science, Zhejiang University
Sep 2023 - Jun 2028 (Expected)

B.S. in Software Engineering, Zhejiang University
Sep 2019 - Jun 2023

experience

Intern, Ant Group
June 2024 - Nov 2025

Research Intern, Alibaba DAMO Academy
June 2023 - June 2024

selected honors

National Scholarship, 2021

Zhejiang Provincial Government Scholarship, 2020

First-class Scholarship of Zhejiang University, 2022, 2021, 2020