CV

Education

• University of Virginia, Charlottesville, VA, USA
  • Ph.D. Candidate in Electrical Engineering (Expected May 2026)
  • Jan. 2025 - Present
• Pennsylvania State University, State College, PA, USA
  • Ph.D. Student in Electrical and Electronics Engineering
  • Sep. 2021 - Dec. 2024
• University of Pennsylvania, Philadelphia, PA, USA
  • M.S.E. in Electrical Engineering
  • Aug. 2019 - May 2021
• Nanjing University, Nanjing, Jiangsu, China
  • B.S. in Electronic Information Science and Technology
  • Sep. 2015 - Jun. 2019

Research Interests

• Large Language Models (LLMs): Transformers, In-context learning (ICL), Chain-of-Thought (CoT)
• Reinforcement Learning: RLHF, Personalized RLHF, Low-Rank Adaptation (LoRA)
• Distributed Learning: Federated Learning, Representation Learning, Heterogeneity

Selected Publications

Conference Proceedings

• [NeurIPS 2025] Renpu Liu, Jing Yang. “Unlabeled Data Can Provably Enhance In-Context Learning of Transformers.” The 39th Annual Conference on Neural Information Processing Systems.
• [ICLR 2025] Renpu Liu, Ruida Zhou, Cong Shen, and Jing Yang. “On the Learn-to-Optimize Capabilities of Transformers in In-Context Sparse Recovery.” The 13th International Conference on Learning Representations.
• [AISTATS 2025] Renpu Liu, Peng Wang, Donghao Li, Cong Shen, Jing Yang. “A Shared Low-Rank Adaptation Approach to Personalized RLHF.” The 28th International Conference on Artificial Intelligence and Statistics.
• [ICML 2024] Renpu Liu, Cong Shen, and Jing Yang. “Federated Representation Learning in the Under-Parameterized Regime.” The 41st International Conference on Machine Learning.
• [ISIT 2025] Renpu Liu, Liwen Zhong, Wooram Lee, and Jing Yang. “In-Context Learning Based Efficient Spectrum Sensing.” IEEE International Symposium on Information Theory.
• [ISIT 2023] Renpu Liu, Jing Yang, and Cong Shen. “Exploiting Feature Heterogeneity for Improved Generalization in Federated Multi-task Learning.” IEEE International Symposium on Information Theory.
• [ISIT 2021] Xingran Chen, Renpu Liu, Shaochong Wang, and Shirin Saeedi Bidokhti. “Timely Broadcasting in Erasure Networks: Age-Rate Tradeoffs.” IEEE International Symposium on Information Theory.

Under Review

• [ICASSP 2026] Li Fan, Zhoubin Kou, Renpu Liu, Jing Yang and Cong Shen. “Augmented In-Context Learning for Wireless Symbol Detection.”
• [ICASSP 2026] Zhoubin Kou, Renpu Liu, Cong Shen and Jing Yang. “Multi-task Transformer-Based Receiver for OFDM Channel Estimation and Symbol Detection.”

Research Experience

University of Virginia / Pennsylvania State University Research Assistant (Advisor: Prof. Jing Yang) Sep. 2021 - Present

• In-Context Learning & Optimization: Proved that unlabeled data enhances ICL performance and developed a framework where Transformers emulate iterative refinement (CoT). Demonstrated that Transformers can implement LISTA-type learning-to-optimize algorithms for sparse recovery.
• Personalized RLHF: Proposed a framework using shared Low-Rank Adaptation (LoRA) to capture user-specific preferences while reducing sample complexity.
• Federated Learning: Developed algorithms for Federated Representation Learning in under-parameterized regimes and exploited feature heterogeneity to improve generalization.

University of Pennsylvania Research Assistant (Advisor: Prof. Shirin Saeedi Bidokhti) Aug. 2019 - May 2021

• Information Theory: Investigated Age of Information (AoI) and rate tradeoffs in broadcast channels.
• Developed simulation scripts to analyze coding policies.