Hieu Trung Nguyen

Hieu Trung Nguyen

SEEM PhD Student @ CUHK

About me

I am 2nd year PhD student at The Chinese University of Hong Kong (CUHK), advised by Prof. Viet Anh Nguyen. Prior to my PhD, I spent 2 wonderful years at VinAI Research as a Research Resident.

While I have broad experience in recommender systems, graph neural networks, and continual learning, my current research focuses on reasoning optimization for Transformer-based Language Models. I work to improve LLM reasoning performance, diversity, and efficiency using techniques like LLM Post-Training (GRPO, self-distillation), KV Cache Compression, model pruning, and routing.

I am currently open to internship opportunities and would love to connect. Please feel free to reach out via email (hilljun.2000@gmail.com) or WeChat (ID: junhill9961).

📜 Publications

  1. Adaptive Rollout Allocation for Online Reinforcement Learning with Verifiable Rewards, with Bao Nguyen, Wenao Ma, Yuzhi Zhao, Ruifeng She and Viet Anh Nguyen. ICLR 2026 - Paper - Code

  2. Reasoning Planning for Language Models, with Bao Nguyen, Ruifeng She, Xiaojin Fu and Viet Anh Nguyen. NeurIPS 2025 (Spotlight) - Paper / Code

  3. Mixture-of-Personas Language Models for Population Simulation, with Ngoc Bui, Shantanu Kumar, Julian Theodore, Weikang Qiu and Viet Anh Nguyen and Rex Ying. ACL 2025 (Findings) - Paper

  4. Structured Pruning for Diverse Best-of-N Reasoning Optimization, with Bao Nguyen and Viet Anh Nguyen. ACL 2025 (Findings) - Paper / Code

  5. Task-driven Layerwise Additive Activation Intervention, with Bao Nguyen, Binh Nguyen and Viet Anh Nguyen. NAACL 2025 (Main) - Paper

  6. Cold-start Recommendation by Personalized Embedding Region Elicitation, with Duy Nguyen, Khoa Doan and Viet Anh Nguyen. UAI 2024. - Paper

  7. Explaining Graph Neural Networks via Structure-aware Interaction Index, with Ngoc Bui, Viet Anh Nguyen and Rex Ying. ICML 2024. - Paper / Code

  8. Generative Conditional Distributions by Neural (Entropic) Optimal Transport, with Bao Nguyen, Binh Nguyen, and Viet Anh Nguyen. ICML 2024. - Paper / Code

  9. Retrospective Feature Estimation for Continual Learning, with Nghia D. Nguyen, Ang Li, Hoang Pham, Viet Anh Nguyen, Khoa D. Doan. Transactions on Machine Learning Research (Featured Certification, J2C Certification) - Paper / Code

  10. Combining Soft-Actor Critic with Cross-Entropy Method for Policy Search in Continuous Control, with Khang Tran and Ngoc Hoang Luong. IEEE CEC 2022 (Oral)

📜 Academic Services

I served as a reviewer for some reputable conferences: ICML2026, ICLR2026, ICML2025, ICLR2025, WWW2025, NeurIPS2024.

🏵️ Honors and Awards

  • July 2024: ICLR2026 Travel Grant! (Rio de Janeiro, Brazil)
  • July 2024: ICML2024 Travel Grant! (Vienna, Austria)
  • April 2022: Bachelors Thesis with highest score.
  • September 2018 - April 2022: Honor Student Scholarship for all Academic Years - UIT