<aside>
👋 Hi there, I'm Yixia Li, a prospective Ph.D. candidate at SUSTech. My research interest mainly focuses on Natural Language Processing and Multi-Modal.
</aside>
Contact
Email | Linkedin | Google Scholar | Github
📰 News
- [2024.06.18] Our paper "SeTAR: Out-of-Distribution Detection with Selective Low-Rank Approximation" is available on arXiv.
- [2024.06.13] Our paper "MiLoRA: Harnessing Minor Singular Components for Parameter-Efficient LLM Finetuning" is available on arXiv.
- [2024.05.16] Our paper "PACIT: Unlocking the Power of Examples for Better In-Context Instruction Tuning" is accepted by ACL 2024 Findings.
- [2024.04.19] Our survey "Cantonese Natural Language Processing in the Transformers Era: A Survey and Current Challenges" is accepted by Journal of Language Resources and Evaluation.
🏫 Education
Southern University of Science and Technology
Sept 2023 - July 2027
Prospective Ph.D. Candidate
The Hong Kong Polytechnic University
Sept 2021 - July 2023
M.Sc. in Data Science with Distinction
- Department of Applied Mathematics
- Rank: Top 5%; GPA: 3.8/4.0
Guangzhou University
Sept 2017 - Jun 2021
B.Sc. in Statistics
- Department of Statistics
- Rank: Top 5%; GPA: 3.7/4.0
👨🏻💻 Research Experiences
PolyU SMART Lab
Research Assistant
Feb 2022 - Jan 2023
Supervisor: Dr. Jing Li
- Automatic Poll Generation for Social Media Posts (Project)
- Task-oriented Dialogue System Development ****(Demo)
Huawei AARC Dept
Research Internship
June 2022 - Aug 2022
Supervisor: Jingyuan Yang
- Extractive Long Document Q&A
- Knowledge-enhanced Dialogue System
- Chinese ASR Entity Correction
📄 Publications and Preprints
SeTAR: Out-of-Distribution Detection with Selective Low-Rank Approximation
<aside>
🔍 SeTAR, a training-free method that improves out-of-distribution detection by post-hoc modification of weight matrices, with extended fine-tuning version SeTAR+FT.
</aside>
- arXiv Preprint (2024.06.18) [PDF] [Code]
- Yixia Li, Boya Xiong, Guanhua Chen, Yun Chen