DU Yiming
Email: ydu@se.cuhk.edu.hk
Room 615, ERB
The Chinese University of Hong Kong
Shatin, New Territories, Hong Kong, China
Hi, I am Yiming (Elvin) DU. I am a PhD candidate at the Chinese University of Hong Kong, pursuing my PhD in the Department of SEEM under the supervision of Professor Kam-Fai Wong. I conduct my research at the MoE Key Laboratory of High Confidence Software Technologies (MoE Lab).
From Jan. 2025 to June 2025, I was a visiting researcher at EdinburghNLP, working with Prof. Jeff Z. Pan. I have also been a research intern at Huawei Noah’s Ark Lab (mentored by Baojun Wang), SenseTime Research (mentored by Fei Tan), and previously did research stints at Imperial College London (with Prof. Lucia Specia) and City University of Hong Kong (Web Mining Lab). Prior to CUHK, I obtained my M.Eng. and B.Eng. in Computer Science from Southeast University.
My research lies at the intersection of memory, dialogue, and LLM agents. My long-term goal is to build personal AI assistants that can remember, reason over time, and self-evolve through interaction.
Research Interests
-
Memory for LLM Agents. Long-term & parametric memory, memory operations, retrieval-augmented and reinforcement-learning–based memory reasoning for multi-session agents.
Representative works: MEMORY-T1 (ICLR 2026), MemGuide (AAAI 2026), PerLTQA (Best Paper, SIGHAN 2024), Rethinking Memory in AI (Survey). -
Dialogue Systems & Personal Assistants. Multi-turn dialogue modeling, personalization, multi-modal & multi-session conversation.
Representative works: ReSURE (EMNLP 2025), Sticker Retrieval (AAAI 2025), Flexibly Utilize Memory (EMNLP 2025). -
Data Selection for LLM Training. High-quality multi-turn data curation, noise-robust SFT/RL training pipelines for Edge LLMs.
Representative work: ReSURE (EMNLP 2025).
Honors & Awards
- 2024 — Overseas Research Attachment Programme (ORAP) Award, CUHK.
- 2024 — Best Paper Award, ACL-SIGHAN 2024 (for PerLTQA).
- 2020 — 1st Place, Jiangsu Postgraduate Mathematical Modeling Innovation Practice Competition.
- 2018 – 2020 — Excellent Scholarships: Shenzhen Stock Exchange, Southeast University, Huawei, Keyuan Automation.
- 2018 — 1st Place, Chinese Undergraduate Computer Design Contest; Meritorious Winner, Mathematical Contest in Modeling (MCM).
Academic Services
- Reviewer: ACL, EMNLP, IJCAI, Knowledge-Based Systems (KBS), etc.
- Patent: A dialogue memory storage, extraction and electronical device — Baojun Wang, Yiming Du, Kam-Fai Wong (Huawei, 2024).
news
No news so far...
selected publications
-
MEMORY-T1: Reinforcement Learning for Temporal Reasoning in Multi-Session AgentsIn International Conference on Learning Representations (ICLR), 2026 -
MemGuide: Intent-Driven Memory Selection for Goal-Oriented Multi-Session LLM AgentsIn Proceedings of the AAAI Conference on Artificial Intelligence, 2026Accepted, Oral. Co-first author with Bingbing Wang. -
ReSURE: Regularizing Supervision Unreliability for Multi-turn Dialogue Fine-tuningIn Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2025 -
Flexibly Utilize Memory for Long-Term Conversation via a Fragment-then-Compose FrameworkIn Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2025Accepted, Oral. -
Rethinking Memory in AI: Taxonomy, Operations, Topics, and Future DirectionsarXiv preprint arXiv:2505.00675, 2025 -
MemLens: Benchmarking Multimodal Long-Term Memory in Large Vision-Language ModelsarXiv preprint arXiv:2605.14906, 2025 -
A New Formula for Sticker Retrieval: Reply with Stickers in Multi-Modal and Multi-Session ConversationIn Proceedings of the AAAI Conference on Artificial Intelligence, 2025Co-first author. -
A Comprehensive Study of Multilingual Confidence Estimation on Large Language ModelsIn Findings of the Association for Computational Linguistics: ACL 2025, 2025 -
PerLTQA: A Personal Long-Term Memory Dataset for Memory Classification, Retrieval, and Fusion in Question AnsweringIn Proceedings of the 10th SIGHAN Workshop on Chinese Language Processing (SIGHAN-10), 2024Best Paper Award.