I am a Ph.D. student in the Machine Learning Center at Georgia Tech (ML@GT), advised by Prof. Chao Zhang. I am also fortunate to work with Prof. Le Song. My research interest primarily lies in Machine Learning and Natural Language Processing.
Before that, I obtained my bachelor's degree from Zhejiang University, and I spent my senior year as a visiting student researcher at Harvard Medical School.
- [--- Top ---] Join the Journey: I am currently diving into exciting projects about Large Language Models (LLM), seeking collaborators/mentees with expertise in this field! Computational resources and preliminary validated ideas provided. Contact me if interested in collaboration!
- [Dec. 2023] I will join Microsoft Research as a research intern in Spring 2024, exploring Mixture of Experts for LLMs.
- [May 2023] One paper accepted to KDD'23, we proposed an iterative and adaptive framework for boosting in weakly-supervised learning settings.
- [May 2023] Our paper PTLoss is available on arXiv, we proposed a novel knowledge distillation objective by perturbing the standard KL loss.
- [May 2023] Two papers accepted to ACL'23, discussing cold-start data selection for few-shot LM finetuning and retrieval enhanced LM.
- [Mar. 2023] I will be back to Google Reseach NYC as a student researcher this summer.
- [May 2022] One paper accepted to KDD'22, discussing adaptive multi-view rule discovery.
- [Apr. 2022] One paper accepted to NAACL'22, discussing uncertainty-based active self-training.
- [Mar. 2022] I will join Google Reseach as a student researcher this summer, see you in New York City.
- [Feb. 2022] One paper accepted by ACL'22, discussing interactive weakly-supervised learning.
Education
My current research, revolving around large language models (LLM), is dedicated to more efficient and performant learning with limited supervision.
In particular, I explore novel learning paradigms, in which humans and machines interact more effectively.
I also investigate the learning objective design in order to improve the model trained with noisy/limited supervision.
I am also interested in how to strategically select data to maximize annotation efficiency.
My research thrusts are as follows:
- Large Language Model: Mechanism, Distillation, Improvement of Training Pipeline
- Learning Paradigm: Interactive and Iterative Methods, Weakly-Supervised Learning
- Learning Objective: Loss Design for Calibration/Deviation-Reduction
- Data Selection: Active Learning, Uncertainty Estimation and Propagation
Rongzhi Zhang, Jiaming Shen, Tianqi Liu, Jialu Liu, Michael Bendersky, Marc Najork, and Chao Zhang.
Do Not Blindly Imitate the Teacher: Using Perturbed Loss for Knowledge Distillation
Preprint on arXiv, 2023.
[BibTex] [arXiv]
Rongzhi Zhang, Yue Yu, Jiaming Shen, Xiquan Cui and Chao Zhang.
Local Boosting for Weakly-Supervised Learning
In ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2023.
[BibTex] [arXiv]
Yue Yu, Rongzhi Zhang, Ran Xu, Jieyu Zhang, Jiaming Shen and Chao Zhang.
Cold-Start Data Selection for Few-shot Language Model Fine-tuning:
A Prompt-Based Uncertainty Propagation Approach
In Annual Meeting of the Association for Computational Linguistics (ACL), 2023.
[BibTex] [arXiv]
Yue Yu, Yuchen Zhuang, Rongzhi Zhang, Yu Meng, Jiaming Shen and Chao Zhang.
Zero-Shot Text Classification by Training Data Creation with Progressive Dense Retrieval
In Findings of Annual Meeting of the Association for Computational Linguistics (ACL), 2023.
[BibTex] [arXiv]
Rongzhi Zhang, Yue Yu, Shetty Pranav, Le Song and Chao Zhang.
PRBoost: Prompt-Based Rule Discovery and Boosting for Interactive Weakly-Supervised Learning
In Annual Meeting of the Association for Computational Linguistics (ACL), 2022.
[BibTex] [arXiv]
Rongzhi Zhang, Rebecca West, Xiquan Cui and Chao Zhang.
Adaptive Multi-view Rule Discovery for Weakly-Supervised Compatible Products Prediction
In ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD), 2022.
[BibTex] [arXiv]
Yue Yu, Lingkai Kong, Jieyu Zhang, Rongzhi Zhang and Chao Zhang.
AcTune: Uncertainty-Aware Active Self-Training for Active Fine-Tuning of Pretrained Language Models
In Annual Conference of the North American Chapter of the Association for Computational Linguistics (NAACL), 2022.
[BibTex] [arXiv]
Rongzhi Zhang, Yue Yu and Chao Zhang.
SeqMix: Augmenting Active Sequence Labeling via Sequence Mixup
In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), 2020.
[BibTex] [Slides]
Experiences
- Research Intern | Jan. 2024 - May 2024
Microsoft Research, Redmond, WA
Host:
Shuohang Wang,
Yelong Shen,
Weizhu Chen
- Student Researcher | May 2023 - Dec. 2023
Google Research, New York City, NY
Host:
Jiaming Shen,
Tianqi Liu,
Jialu Liu
- Student Researcher | May 2022 - Dec. 2022
Google Research, New York City, NY
Host:
Jiaming Shen,
Tianqi Liu,
Michael Bendersky
Teaching
- 2023 KDD Student Travel Award, ACM SIGKDD
- 2019 Outstanding Graduate, Zhejiang University
- 2019 Outstanding Undergrad Thesis, Zhejiang University
- 2018 Zhejiang Provincial Government Scholarship (Top 3%)
- 2018 First-class Excellent Undergraduate Scholarship, Zhejiang University (Top 3%)
- 2018 First-class Academic Scholarship, Zhejiang University (Top 3%)
- 2016 Student Innovation Research Funding, Zhejiang Province
- 2014 The first prize in the 31st Chinese Physics Olympiad (CPhO)
- Program Committee / Reviewer: TKDE 2020, 2021; EMNLP 2022; ACL 2022; KDD 2023; NeurIPS 2023, ACL Rolling Review 2023, 2024, ICLR 2024.
- I was a player of Zhejiang University Varsity Men's basketball team, competing in CUBA Division II.