Scholar
Hu Xu
Google Scholar ID: SaH2yWMAAAAJ
Meta AI (FAIR Labs)
Citations & Impact
Citations
23,713
H-index
34
i10-index
50
Publications
20
Co-authors
12
list available
Publications
Resume (English only)
Academic Achievements
- First author of the ICLR Spotlight paper 'Demystifying CLIP Data (MetaCLIP)' in 2024; core contributor to 'The Llama 3 Herd of Models' project; led the development of 'MoDE: CLIP Data Experts via Clustering', presented at CVPR 2024.
Research Experience
- Serves as a research scientist at Facebook AI Research (FAIR). During his Ph.D., he interned at Facebook AI, Amazon AI, and WeChat AI lab. He was also the winner of the Yelp dataset challenge.
Education
- Received a master's degree in microelectronics from Peking University; obtained a Ph.D. in computer science from the University of Illinois at Chicago, advised by Prof. Philip S. Yu and Prof. Bing Liu.
Background
- Research interests include multi-modal pre-training (CLIP and unCLIP), self-supervised learning, and Internet-scale data curation. Specializes in Lifelong Representation Learning, Open-world Learning, Sentiment Analysis, Question Answering, and Conversational AI.
Miscellany
- Open to giving invited talks on insights into his research; his team is looking for research interns in multi-modal pretraining; defended his Ph.D. thesis in February 2020 and open-sourced it; submitted a Ph.D. proposal on lifelong representation learning for NLP in September 2019.
