Dr. Guangyue Xu(许广跃)

I am a final-year Machine Learning(ML) and Natural Language Processing(NLP) Ph.D. student at Michigan State University, jointly supervised by Prof. Parisa Kordjamshidi and Prof. Joyce Y. Chai. My research focus on pre-training large vision-language models and enhance their generalization abilities.

I received my B.E. in Software Engineering from Jilin University and M.S. in Computer Science from Tsinghua University. I also interned in MSRA's Web Search and Mining Group previously.

I will graduate in spring 2024 and I am looking for full-time LLM scientist or engineering jobs!

Email / CV / Google Scholar / Github /

Publications

	GIPCOL: Graph-Injected Soft Prompting for Compositional Zero-Shot Learning Guangyue Xu, Parisa Kordjamshid, Joyce Chai WACV, 2024 project page / arXiv We introduce a GNN into soft-prompting design to improve CLIP's compositional ability.
	MetaReVision: Meta-Learning with Retrieval for Visually Grounded Compositional Concept Acquisition Guangyue Xu, Parisa Kordjamshid, Joyce Chai EMNLP-Finding, 2023 project page / arXiv We meta-train vision-language models using retrieved items to obtain more generalizable token representations and improve vision-language model's compositional ability.
	Prompting large pre-trained vision-language models for compositional concept learning Guangyue Xu, Parisa Kordjamshid, Joyce Chai arXiv, 2022 We systematically investigate various prompting techniques for CLIP in compositional zero-shot learning.
	Zero-Shot Compositional Concept Learning Guangyue Xu, Parisa Kordjamshid, Joyce Chai MetaNLP@ACL, 2021 arXiv We propose a multi-modality reconstruction-based network for RGBD AD, which eliminate the usage of memory bank and pretrained model. Moreover, the proposed method obtains the best trade-off between the accuracy and inference speed.
	Language to Action: Towards Interactive Task Learning with Physical Agents Joyce Chai, Qiaozi Gao, Lanbo She, Shaohua Yang, Sari Saba-Sadiya, Guangyue Xu IJCAI-ECAI (Invited Paper), 2018 arXiv Thoughts and positions about the importance of language communication in human learning and knowledge acquisition.
	Tracking You Through DNS Traffic: Linking User Sessions By Clustering With Dirichlet Mixture Model Guangyue Xu, Mingxuan Sun, Junjie Zhang, Dae Wook Kim MSWiM, 2017 arXiv We provide a Bayesian nonparametric model, constrained Dirichlet multinomial mixture (CDMM), to analyze user behavior based on the Domain Name System(DNS) data.

Service and Activities

Reviewer: EACL 2021, EMNLP 2023, ACM MM 2023, EACL 2023, ACL 2023

Teaching

CSE 102: Algorithmic Thinking and Programming(Spring 2020, Fall 2020, Fall 2021) - Michigan State University - Teaching Assistant
CSE 231: Introduction to Programming I(Spring 2022, Fall 2022) - Michigan State University - Teaching Assistant
CSE 232: Introduction to Programming II(Spring 2023) - Michigan State Univesity - Teaching Assistant
CSE 440: Introduction to Artificial Intelligence(Fall 2019, Spring 2021) - Michigan State University - Teaching Assistant

Website template from JonBarron.