E: yichaocaiadelaideeduau
AIML, University of Adelaide
LG.25.04, AIML Building
Cnr North Tce & Frome Rd
Adelaide SA 5000, Australia
0000-0003-1607-8948
Yichao Cai is a third-year Ph.D. student at the
Australian Institute for Machine Learning (AIML),
University of Adelaide, advised by Prof.
Javen Qinfeng Shi, Dr
Zhen Zhang, and Dr
Yuhang Liu.
Before beginning his Ph.D., he spent four years in industry. He received his M.S. from
Wuhan University of Technology (WUT), advised by Prof. Xiao Zhou, and earned his B.E. from WUT as well. During his master's studies, he was a visiting student researcher at
Berkeley DeepDrive, UC Berkeley, advised by Dr
Ching-Yao Chan.
His current research centers on multimodal representation learning and principled approaches to disentanglement, with a particular emphasis on identifiability analysis, causal modeling, and leveraging language as an inductive signal. Looking ahead, he aims to develop principled AI systems capable of reliably uncovering and representing the semantic structures underlying complex data.
- 14 Apr 2025: Check out our new preprint—"On the Value of Cross-Modal Misalignment in Multimodal Representation Learning".
- 02 Jul 2024: Our work,"CLAP: Isolating Content from Style through Contrastive Learning with Augmented Prompts", is accepted to appear at ECCV 2024.
- Semester 2, 2025, Head Tutor — Statistical Machine Learning @ University of Adelaide
- Trimester 2, 2025, Teaching Assistant — Using Machine Learning Tools @ University of Adelaide
- Trimester 1, 2025, Teaching Assistant — Concepts in Artificial Intelligence and Machine Learning @ University of Adelaide
- [1] Yichao Cai, Yuhang Liu, Zhen Zhang, and Javen Qinfeng Shi. CLAP: Isolating Content from Style through Contrastive Learning with Augmented Prompts. In Computer Vision – ECCV 2024: 18th European Conference, Milan, Italy, September. [ abs | pdf | project page ]
- [2] Yuhang Liu, Dong Gong, Yichao Cai, Erdun Gao, Zhen Zhang, Biwei Huang, Mingming Gong, Anton van den Hengel, and Javen Qinfeng Shi. I Predict Therefore I Am: Is Next Token Prediction Enough to Learn Human-Interpretable Concepts from Data? Preprint. Under review. [ abs | pdf ]
- [1] Yichao Cai*, Yuhang Liu*, Erdun Gao, Tianjiao Jiang, Zhen Zhang, Anton van hen Hengel, and Javen Qinfeng Shi (* Equal Contributions). On the Value of Cross-Modal Misalignment in Multimodal Representation Learning. Preprint. Under review. [ abs | pdf | project page ]