I am currently a second-year PhD student at ASI Lab, National University of Singapore, mentored by Prof. Dianbo Liu.
My research interests lie in autonomous ML agents (especially AI Scientist), generative models, and representation learning, which I believe have the potential to truly transform how science is done.
Before that, I was a research assistant at BNRist, Tsinghua University, mentored by Prof. Xiangyang Ji and Prof. Yi Xu. I received my master's degree in 2022 from the Department of Automation, Tsinghua University, under the supervision of Prof. Xiangyang Ji. My previous research has focused on unsupervised segmentation and text-to-motion synthesis, aiming to reduce human annotation effort and improve learning efficiency.
Research Interests
- Autonomous ML Agents & AI Scientist
- Representation Learning
- Generative Models
Previously: Text-to-Motion Synthesis, Unsupervised Segmentation
News
- [2025] New preprint: FML-bench — a benchmark for evaluating autonomous ML research agents.
- [2025] New preprint: LanSE — a method for human-like analysis of AI-generated images using language-grounded sparse encoders.
- [2024] Our paper on representation collapsing problems in vector quantization is accepted to NeurIPS Safe Generative AI Workshop 2024.
- [2024] Our paper ParCo on part-coordinating text-to-motion synthesis is accepted to ECCV 2024.
- [Dec. 2022] Our paper on unsupervised foreground-background segmentation is accepted to AAAI 2023 (Main Track, Oral).
- [Oct. 2021] Our paper on learning foreground-background segmentation from improved layered GANs is accepted to WACV 2022.
Publications
Services
Conference Reviewer: ICLR'25, AAAI'26, ICLR'26, ICML'26
Visitors
