Ph.D Student, Computer Science
Nanyang Technological University, Singapore
I am a first-year PhD student and luckily advised by the brilliant and kind researcher Prof. Ziwei Liu. With his guidance, I am passionate about deep learning research in the following topics:
(1) Pathways system and PaLM: a next-generation AI model that can multi-task across domains and modalities.
(2) Embodied AI: an autonomous agent that learns to solve challenging tasks within environment through interaction and exploration.
These are moonshot wild dreams and also what I will focus on in the long run.
Currently I base my first step research topics on computer-vision-in-the-wild and emerging abilities in large vision language models (e.g. Chain of Thought, Multi-step Reasoning).
If you are an undergraduate & master student and want to dive in these topics, feel free to drop an email~
I have been fortunately doing research at/with
Nov. 2022 - Present: Microsoft Research, Redmond
Explore large models emerging abilities and deep mysteries of intelligence.
Sep. 2020 - Dec. 2021: Microsoft Research, Shanghai
Beautiful and relaxing WestBud office, explore the unknowns with chill and smart colleagues.
Oct. 2019 - Aug. 2020 (remote till May 2021): Berkeley AI Research, CA, USA
Sunny summer and warm winter. Go Cal and Roll on your Golden Bears!
May 2020 - Dec. 2021: Prof. Han Zhao from UIUC, USA
Learn to write a paper with machine learning taste.
May 2018 - Oct. 2019: DiDi Visual Perception Team, Beijing
First internship and two papers there.
- Sparse Mixture-of-Experts are Domain Generalizable Learners
[code]Short version in NeurIPS 2022 Workshop on Distribution Shift.
- Invariant information bottleneck for domain generalization
AAAI 2022,In Proceedings of the AAAI Conference on Artificial Intelligence. [code]
- Energy-Based Open-World Uncertainty Modeling for Confidence Calibration
ICCV 2021,In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV). [code]
- Learning invariant representations and risks for semi-supervised domain adaptation
CVPR 2021,In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. [code]
- MADAN: multi-source adversarial domain aggregation network for domain adaptation
IJCV 2021,International Journal of Computer Vision.
- Rethinking distributional matching based domain adaptation
- Multi-source domain adaptation for semantic segmentation
NeurIPS 2019,In Neural Information Processing Systems. [code]
Conference Reviewer / Program Committee:
ICCV (2021), NeurIPS (2022), BMVC (2023), AAAI (2023), CVPR (2023), AISTATS (2023).
Journal Reviewer: Pattern Recognition.
Slab@NTU: GPU Server Management (~70 users, 400+ GPUs)
Awards and Fellowships
- AISG Fellowship 2022
- Microsoft Starbridge Program 2020
- Silver Medal, China Collegiate Programming Competition (CCPC), ACM Regional, 2017
- Deadlift: 100kg, 1-3RM
- Squat: 100kg, 1-3RM
- Bench Press: 80kg, 3RM
- 5KM Run: ~34 min
- Love watching a lot of Youtube videos and reading daily world news.
- Enjoy participating meaningful social activities (connect with community).
- Dream to be a film director.