Yasi Zhang

PhD Student@UCLA. Generative AI, Multimodality, Computer Vision,and Unsupervised Learning.

prof_pic.png

🌈 Greetings! I am Yasi Zhang (She/They), currently a PhD student in the Department of Statistics and Data Science at UCLA. I’m super fortunate to be co-advised by two of the best advisors in the world, Prof. Ying Nian Wu and Prof. Oscar Leong. Previously, I earned my bachelor’s degree in Data Science and Big Data Technology from Fudan University. I conducted an Applied Scientist internship at Amazon AWS AI Labs in Summer 2024 and a Student Researcher internship at Google Research in Spring and Summer 2025.

My current research goal is to build a unified foundation for powerful multimodal generative models capable of understanding, reasoning, aligning, and generating high-dimensional data across diverse modalities.

✨ I’m actively seeking research internship opportunities for Summer 2026.

News

Oct 3, 2025 This new book chapter systematically examined (almost) all the methods in the field of learned regularization + inverse problems.
Sep 23, 2025 What truly defines the success of an image editor? The answer is in 🤖EdiVal-Agent.
Jun 5, 2025 If you haven’t done so, please check out our new papers on Denoising Score Distillation and Restoration Score Distillation, both contributing to a principled framework for learning generative models from corrupted data.
Jan 22, 2025 One paper on lifelong reinforcement learning accepted to AISTATS 2025.
Oct 11, 2024 Honored to receive the NeurIPS 2024 Scholar Award. See you in Vancouver.
Sep 25, 2024 One paper on flow matching priors for inverse problems was accepted to NeurIPS 2024.
Jul 1, 2024 Two papers on diffusion models were accepted to ECCV 2024. One is on text-to-image alignment in stable diffusion, and the other is on generalization.

Selected Publications

  1. Book Chapter
    Learning Regularization Functionals for Inverse Problems: A Comparative Study
    Johannes Hertrich*, Hok Shing Wong*, Alexander Denker, Stanislas Ducotterd, Zhenghan Fang, Markus Haltmeier, Željko Kereta, Erich Kobler, Oscar Leong, Mohammad Sadegh Salehi, Carola-Bibiane Schönlieb, Johannes Schwab, Zakhar Shumaylov, Jeremias Sulam, German Shâma Wache, Martin Zach, Yasi Zhang, Matthias J. Ehrhardt**, and Sebastian Neumayer**
    2025
    PDF Code
  2. Preprint
    EdiVal-Agent: An Object-Centric Framework for Automated, Scalable, Fine-Grained Evaluation of Multi-Turn Editing
    Tianyu Chen*, Yasi Zhang*, Zhi Zhang, Peiyu Yu, Shu Wang, Zhendong Wang, Kevin Lin, Xiaofei Wang, Zhengyuan Yang, Linjie Li, Chung-Ching Lin, Jianwen Xie, Oscar Leong**, Lijuan Wang**, Ying Nian Wu**, and Mingyuan Zhou**
    2025
    PDF Website
  3. Preprint
    Restoration Score Distillation: From Corrupted Diffusion Pretraining to One-Step High-Quality Generation
    Yasi Zhang*, Tianyu Chen*, Zhendong Wang, Ying Nian Wu, Mingyuan Zhou**, and Oscar Leong**
    2025
    PDF
  4. Preprint
    Denoising Score Distillation: From Noisy Diffusion Pretraining to One-Step High-Quality Generation
    Tianyu Chen*, Yasi Zhang*, Zhendong Wang, Ying Nian Wu, Oscar Leong**, and Mingyuan Zhou**
    2025
    PDF
  5. Preprint
    Learning Difference-of-Convex Regularizers for Inverse Problems: A Flexible Framework with Theoretical Guarantees
    Yasi Zhang, and Oscar Leong
    2025
    PDF
  6. AISTATS
    Statistical Guarantees for Lifelong Reinforcement Learning using PAC-Bayesian Theory
    Zhi Zhang, Chris Chow, Yasi Zhang, Yanchao Sun, Haochen Zhang, Eric Hanchen Jiang, Han Liu, Furong Huang, Yuchen Cui, and Oscar Hernan Madrid Padilla
    In International Conference on Artificial Intelligence and Statistics 2025
    PDF
  7. NeurIPS
    Flow Priors for Linear Inverse Problems via Iterative Corrupted Trajectory Matching
    Yasi Zhang, Peiyu Yu, Yaxuan Zhu, Yingshan Chang, Feng Gao, Ying Nian Wu, and Oscar Leong
    In Neural Information Processing Systems 2024
    PDF
  8. ECCV
    Object-Conditioned Energy-Based Attention Map Alignment in Text-to-Image Diffusion Models
    Yasi Zhang, Peiyu Yu, and Ying Nian Wu
    In European Conference on Computer Vision 2024
    PDF
  9. ECCV
    Skews in the Phenomenon Space Hinder Generalization in Text-to-Image Generation
    Yingshan Chang, Yasi Zhang, Zhiyuan Fang, Yingnian Wu, Yonatan Bisk, and Feng Gao
    In European Conference on Computer Vision 2024
    PDF

Education

  • University of California, Los Angeles, 2022.09 - Present
    PhD in Statistics and Data Science, Most Promising Statistician Award
  • Fudan University, 2018.09 - 2022.06
    B.S. in Data Science and Big Data Technology, First-Class Honor

Experience

  • Google Research, Student Researcher, Spring&Summer 2025
  • Amazon AWS AI Labs, Applied Scientist Intern, Summer 2024