Rongyi Zhu

I am a Ph.D student in the Department of Computer Science at Stony Brook University, advised by Prof. Ting Wang.

I received my M.S. degree in Computer Science from the University of Rochester in 2024, where I worked with Prof. Chenliang Xu in trustworthy AI.

Email / Twitter / LinkedIn / Github / Scholar

Research

My research focuses on understanding and controlling the fundamental mechanisms of deep learning through targeted perturbations.

	Self-Improving Model Steering Rongyi Zhu, Yuhui Wang, Tanqiu Jiang, Jiacheng Liang, Ting Wang arXiv preprint 2025 We introduce a self-improving model-steering framework by iteratively generating contrastive samples.
	Steering Protein Language Models Long-Kai Huang, Rongyi Zhu, Bing He, Jianhua Yao ICML 2025 We propose an activation editing method in Protein Language Model to steer the sequence generation toward desired properties and its downstream application to protein optimizaiton.
	Learning to Transform Dynamically for Better Adversarial Transferability Rongyi Zhu, Zeiliang Zhang, Susan Liang, Zhuo Liu, Chenliang Xu CVPR 2024 We conceptualize the selection of optimal transformation combinations in adversaril attack as a trajectory optimization problem
	ST-DPGAN: A Privacy-preserving Framework for Spatio-temporal Data Generation Wei Shao, Rongyi Zhu, Cai Yang, Chandra Thapa, Muhammad Ejaz Ahmed, Seyit Camtepe, Rui Zhang, Duyong Kim, Hamid Menouar, Flora Salim IEEE Internet of Things Journal We design a simplified version of a spatio-temporal attention block to effectively align the spatial and temporal information while providing a privacy guarantee
	Video Understanding with Large Language Models: A Survey Yunlong Tang, Jing Bi, Siting Xu, Luchuan Song, Susan Liang, Teng Wang, Daoan Zhang, Jie An, Jingyang Lin, Rongyi Zhu, Ali Vosoughi, Chao Huang, Zeliang Zhang, Feng Zheng, Jianguo Zhang, Ping Luo, Jiebo Luo, Chenliang Xu IEEE Transactions on Circuits and Systems for Video Technology We examine the unique characteristics of Vid-LLMs and presents a study of the tasks and datasets for Vid-LLMs
	Differentially Private Low-Rank Adaptation of Large Language Model Using Federated Learning Xiao-yang Liu, Rongyi Zhu, Daochen Zha, Jiechao, Gao, Shan Zhong, Meikang Qiu IEEE transcation on Management Information Systems We investigate how data privacy can be ensured in LLM fine-tuning through practical federated learning approaches.

Experience
AI for science research, Tencent AI lab , Shenzhen, China Reserach Intern • April. 2024 to Aug. 2024 Mentor: Dr. Long-kai Huang and Dr. Jianhua Yao
Neural network robustness analysis, University of Rochester , NY, USA Reserach Intern • September. 2023 to April. 2024 Advisors: Prof. Chenliang Xu
Reliable autonomous driving, University of California - Davis , CA, USA Reserach Intern • June. 2023 to August. 2023 Advisors: Prof. Junshang, Zhang
Intelligence Driving Group, Baidu, Inc , Beijing, China Software Engineer Intern • Mar. 2021 to July. 2021 Mentor: Dr. Shaofeng Guo and Dr. Jie Chen

Thanks to Jon Barron for sharing the source code.

Research

Experience