Welcome to my personal website! I am Zhiyuan Liu (Chinese: 刘知远), a highly motivated and passionate undergraduate student majoring in Software Engineering at Harbin Institute of Technology (HIT). I am currently a junior (Class of 2026) with a strong academic record and a deep interest in Data-centric AI, particularly in Data-Efficient Artificial Intelligence.
My research is centered around the burgeoning field of Data-Efficient AI, which tackles the critical challenge of data dependency in modern AI. Current artificial intelligence models often demand vast amounts of training data, leading to prohibitively high training costs, especially for large-scale models. I am deeply interested in exploring innovative approaches to utilize data more efficiently, including techniques for scientific data cleaning and synthesis, ultimately striving towards the goal of truly data-efficient artificial intelligence. My specific research interests directly align with this vision, focusing on:
- Dataset Distillation: Developing methods for efficient and lossless dataset compression, enabling effective training with significantly reduced data.
- Knowledge Distillation: Investigating novel techniques to transfer knowledge from large, data-hungry models to compact student models, promoting efficient knowledge representation and transfer.
You can find my CV here: Zhiyuan Liu’s Curriculum Vitae.
🔥 News
- 2025.03: 🎉 The code of our paper NCFM has been released.
- 2025.02: 🎉 Paper “Dataset Distillation with Neural Characteristic Function: A Minmax Perspective” was accepted by CVPR 2025 (Rating: 5/5/5). Thanks!
- 2025.01: 🤗 We release an open-sourse repo “Awesome Dataset Reduction”, which collects recent awesome dataset reduction papers! Feel free to contribute your suggestions!
- 2024.08: 🤗 Started Research Intern at EPIC Lab, Shanghai Jiao Tong University, focusing on Dataset Distillation and Knowledge Distillation.
📝 Publications
Dataset Distillation with Neural Characteristic Function: A Minmax Perspective
Shaobo Wang, Yicun Yang, Zhiyuan Liu, Chenghao Sun, Xuming Hu, Conghui He, Linfeng Zhang
- Pioneered NCFM, a novel dataset distillation approach that reframes the problem from a Characteristic Function perspective. This innovative approach casts dataset distillation within a min-max framework.
- NCFM achieves state-of-the-art performance while drastically reducing GPU memory requirements to 1/300th of prior leading methods and accelerating training by 20x.
- Project
🎖 Honors and Awards
- 2023.12 National Scholarship (Top 1%), Ministry of Education of China
- 2023.11 Huawei Smart Base Scholarship, Huawei Technologies Co., Ltd.
📖 Educations
- 2022.08 - Present, Harbin Institute of Technology, Bachelor of Software Engineering
💻 Internships
- 2024.08 - Present, Research Intern, EPIC Lab, Shanghai Jiao Tong University (Advised by Prof. Linfeng Zhang)