Scholar
Huanxuan Liao
Google Scholar ID: sRcWOKUAAAAJ
Institute of Automation, Chinese Academy of Sciences
Citations & Impact
Citations
73
H-index
4
i10-index
2
Publications
15
Co-authors
8
list available
Publications
Resume (English only)
Academic Achievements
- Publications: SparK: Query-Aware Unstructured Sparsity with Recoverable KV Cache Channel Pruning (AAAI 2026); Awards: National Scholarship (awarded by Ministry of Education at CASIA in 2025, NCEPU in 2022 & 2021); Outstanding Graduate (awarded by Beijing Ministry of Education and NCEPU in 2023).
Research Experience
- Currently pursuing PhD research focused on efficiency optimization for large language models.
Education
- PhD: University of Chinese Academy of Sciences (UCAS) and Institute of Automation, Chinese Academy of Sciences (CASIA), 2023-present, Pattern Recognition and Intelligent Systems, supervised by Prof. Shizhu He and Prof. Kang Liu; B.Eng.: North China Electric Power University (NCEPU), 2019-2023, Intelligence Science and Technology, supervised by Prof. Min Shi and Prof. Yun Ju.
Background
- Research Interests: At the intersection of natural language processing and machine learning, focusing on large language models (LLMs). Aims to enhance the practicality and accessibility of LLMs in real-world scenarios through optimizing LLMs for efficient deployment and strengthening smaller, more economical models. Research directions include but are not limited to Efficient Reasoning, Long-sequence Alignment and Reasoning, Efficient Attention and Architectures, and Long Video Understanding.
Miscellany
- Personal interests not specifically mentioned.
