KiMI: Knowledge-integrated Machine Intelligence
KiMI is a research lab in HKUST(GZ) dedicated to exploring the intersection of artificial intelligence and domain-specific knowledge. Our mission is to develop intelligent solutions that not only learn from data but also leverage existing information to enhance reasoning and understanding.
Our research primarily focuses on seamlessly integrate vast reservoirs of knowledge with cutting-edge machine learning techniques. Key areas of exploration include:
- Retrieval augmented generation: We explore methods to enhance the generation of LLMs by retrieving from large (multi-modal) corpus.
- LLM reasoning: We concentrate on improving the reasoning ability of large language model (LLM) with reinforcement learning and architecture optimization.
- Downstream LLMs: We focus on applying LLMs to complex problems, thereby transforming traditional workflows and fostering innovation.
- AI for science: We harness the power of AI to address complex scientific challenges, driving innovation and discovery across various scientific disciplines.
Recruitment: Student who has received an RBM Mphil or DCAI-MSc offer from HKUST(GZ) can directly contact me for a discussion for potential supervision and future PhD opportunities.
Main collaborators
- Lei Chen: Chair Professor, HKUST(GZ)
- Quanming Yao: Associate Professor, Tsinghua University
Ph.D. students
- Zhuangfei Hu: start from 2024.09
- Siyi Liu: start from 2025.02
- Hengwei Dai: start from 2025.02 (part-time)
- Yuhang He: start from 2025.09
- Xiaotang Wang: start from 2025.09
- Hange Zhou: start from 2025.09
- Haodong Wu: start from 2026.02
MPhil students
- Runqing Xu: start from 2024.09
- Mingkai Qiu: start from 2024.09
- Hongyu Ge: start from 2025.09
- Xiaorong Zhu: start from 2025.09
- Boyang Zhao: start from 2025.09
- Bairui Zhang: start from 2025.09
RAs
- Enjun Du (2024.07 - ). UG in BIT
- Yijia Li (2025.09 - ). MSc in CUHK-SZ