The HKU Musketeers Foundation Institute of Data Science 香港大學同心基金數據科學研究院

16/06/2026

Kicking Off HKU Summer Research Programme 2026 with IDS ☔️🌈
IDS 「港大 2026 暑期研究項目」正式展開
 
A rainy summer day in Hong Kong marked the beginning of a ten-week research journey at HKU IDS.

Organised by the HKU Graduate School HKU Graduate School, the HKU Summer Research Programme 2026 (SRP2026) runs from 1 June to 7 August. This year, 98 students were selected from 2,082 applications across Faculties to join this immersive summer research experience at HKU.
 
At IDS, six SRP researchers have been admitted this year, coming from City University of Hong Kong, Australian National University, Tsinghua University, University College London, New York University, and HKU. Through one-to-one mentorship with HKU professoriate staff, individual research projects, research skills training, project presentations, teambuilding, ceremonies, networking, and extra-curricular activities, the programme offers students a focused opportunity to explore research life at HKU.
 
At today’s (16 June) orientation, the SRP new-bloods heard practical advice from past SRP awardees on conducting research, studying smart, and crafting impactful presentation posters. They also gathered local “survival hacks”, from navigating Hong Kong’s minibuses to tracking down the best food spots in town.
 
The session concluded with a relaxed ice-breaking activity, setting the tone for a close and supportive SRP community. Over the next ten weeks, we look forward to seeing our young researchers explore new ideas, experience Hong Kong, build friendships, and broaden research horizons that will stay with them beyond the summer.
 
Stay tuned for more moments from the SRP2026 journey with IDS.

雨中的香港夏日，迎來為期十星期的研究旅程。

由港大研究生院主辦的「港大 2026 暑期研究項目」（SRP2026）於 6 月 1 日至 8 月 7 日舉行。今年項目從 2,082 份申請中遴選 98 名學生，參與跨學院暑期研究體驗。

IDS 今年迎來六位 SRP 研究新秀，分別來自香港城市大學、澳洲國立大學、清華大學、倫敦大學學院、紐約大學及香港大學。項目涵蓋教授一對一指導、個人研究、研究技巧培訓、項目匯報、團隊建立、交流及課外體驗，讓同學深入了解港大的研究環境與學術生活。

在今日（6 月 16 日）的迎新活動中，同學從歷屆 SRP 得獎者身上汲取實用經驗，包括如何開展研究、規劃學習，以及設計具說服力的學術海報。活動亦分享香港生活資訊，協助研究員更快投入暑期生活。

迎新活動以破冰環節作結，為暑期研究社群揭開序幕。未來十星期，期待各位研究員在 HKU IDS 探索研究方向、認識香港、建立友誼，並帶走一段難忘的學術旅程。

11/06/2026

📢 HKU IDS Scholar Prof Bo DAI shares his views in the HKU Bulletin on how AI can open new possibilities for creativity, learning, and exploration, drawing on his Common Core course “Gen AI and the Future of Creativity”.

Read the feature:

Two Common Core courses are encouraging students to engage in the debate on whether AI can be creative, one from an arts perspective, the other from engineering.

10/06/2026

Build the Brains Behind the Bots! IDS Summer Workshop 2026
玩轉 AI 大腦 🌞 IDS 數據科學夏令營 2026

This summer, HKU IDS is launching IDS Summer Workshop 2026 – AI Agents: Advances, Challenges, and Future Directions, a three-day on-campus workshop led by Prof Chao HUANG.

Set on the HKU campus, the workshop will introduce participants to the foundations of Large Language Models and AI agents, covering LLM architecture, agent design, planning and reasoning, memory, multi-agent collaboration, real-world applications, and capstone project activities.

The workshop is designed for students at undergraduate level or above who are interested in generative AI, intelligent systems, and future research directions. Basic Computer Science or AI knowledge will be a plus.

🗓️ 19–21 August 2026
🏫 IDS Seminar Room, P603, Graduate House, HKU

⏰ Early Bird Offer: Enjoy 10% OFF when you apply on or before 10 July 2026!
🎓 Previous IDS Summer Institute participants may also enjoy a 10% OFF discount (subject to verification and quota).

✍️ Details and application: https://datascience.hku.hk/summer26fb

今年夏天，HKU IDS 將於港大校園主辦「IDS 數據科學工作坊」。

為期三日的工作坊，將由 黃超教授 帶領，探索大型語言模型 (LLMs) 與 AI Agents 的基礎概念，內容涵蓋 LLM 架構、Agent 設計、規劃與推理、記憶機制、多 Agent 協作、實際應用，以及專題項目。

「IDS 數據科學工作坊」適合對生成式 AI、智能系統及未來研究方向感興趣、學士學位課程或以上程度學生參加，尤其適合具基本電腦科學或 AI 知識者。

🗓️ 8 月 19–21 日
🏫 港大研究生院 P603 講室

⏰ 早鳥優惠：於 2026 年 7 月 10 日或之前報名，即可享有九折優惠！
🎓 曾參加 IDS 暑期學院的學員亦可享九折優惠（須經核實及視乎名額而定）。

✍️ 詳情及報名：https://datascience.hku.hk/summer26fb

08/06/2026

🏛 HKU IDS Distinguished Speaker Series 傑出講者系列 #9 | Prof Jianjun Shi 史建軍教授

HKU IDS is honoured to welcome Prof Jianjun Shi from Georgia Institute of Technology for the next Distinguished Speaker Seminar at HKU.

Prof Shi is the Carolyn J. Stewart Chair and Professor at Georgia Tech, and a Member of the US National Academy of Engineering. A leading scholar in data-driven quality, manufacturing systems, and industrial systems engineering, his work has shaped how data science and system informatics can support the modelling, monitoring, diagnosis, and control of complex industrial systems.

🎙️ Data Modality, Data Science, and Multistage Manufacturing

🗓️ 2 July 2026 (Thursday)
🕓 4:00 pm – 5:00 pm HKT
📍 Inno Wing II, HKU
🔗 Details & registration: https://datascience.hku.hk/s20260702fb

HKU IDS 榮幸邀請美國佐治亞理工學院 Carolyn J. Stewart 講席教授、美國國家工程院院士史建軍教授親臨港大開講。

史教授是數據驅動品質工程、製造系統及工業系統工程領域的國際知名學者，其研究推動數據科學與系統資訊學應用於複雜工業系統的建模、監測、診斷與控制。

🎙️ Data Modality, Data Science, and Multistage Manufacturing

🗓️ 2026 年 7 月 2 日（星期四）
🕓 下午 4 時至 5 時（香港時間）
📍 香港大學 Inno Wing II
🔗 詳情及報名：https://datascience.hku.hk/s20260702fb

06/06/2026

📰 Issue 6 of the HKU IDS e-Newsletter is now available!
第六期《HKU IDS 電子報》正式推出！

This issue offers a curated look at the latest facets of IDS, from global AI governance and interdisciplinary academic exchange to student research engagement, research news, awards, and talent development.

Highlights include:
🗯️ Prof Boris Babic’s post-conference reflection on Hong Kong Global AI Governance Conference 2026
💼 2nd Hong Kong AI and Frontier Technology Forum
🧑‍🏫 3rd Shanghai–Hong Kong AI Forum
🏫 Conference on Parsimony and Learning (CPAL) 2026 scholar and student takeaways
🪙 IDS Research Seed Funds 2025 awarded projects
🆕 Updates from our research postgraduate community

📖 Read the latest issue:
https://datascience.hku.hk/enews6fb

今期內容彙集 IDS 近期多個重要發展面向，由全球人工智能治理、跨學科學術交流，到學生科研參與、研究消息、獎項及人才培育，帶領大家更完整地回顧 IDS 最近動向。

內容重點包括：
🗯️ Boris Babic 教授就「香港全球人工智能治理峰會 2026」作會後分享
💼 第二屆「香港人工智能及前沿科技論壇」
🧑‍🏫 第三屆「滬港 AI 學術交流會」
🏫 Conference on Parsimony and Learning 2026 (CPAL 2026) 師生交流體會
🪙 IDS Research Seed Funds 2025 獲資助項目
🆕 IDS 研究生社群最新發展

📖立即閱讀電子報：
https://datascience.hku.hk/enews6fb

05/06/2026

🌟 Event Recap 活動回顧 | HKU IDS Guest Seminar – Learning to Stabilize Plasma: Provable Imitation Learning for Nuclear Fusion Control 🔬⚛️

A big thank you to Prof Wenlong MOU from the University of Toronto for delivering an insightful HKU IDS Guest Seminar on 5 June 2026 at Inno Wing II, HKU.

The seminar explored how machine learning can contribute to one of the central challenges in practical nuclear fusion: maintaining the stability of magnetically confined plasma. Prof MOU introduced recent advances in provable imitation learning for plasma control, starting from expert controllers for fully observed models and developing feedback policies that operate using experimentally available measurements.

The talk further examined the trade-off between adaptivity and stability in offline and online imitation learning. While offline behaviour cloning can adapt to complex initial distributions, it may suffer from exponential error compounding. Online algorithms, by contrast, offer a route towards long-term stability with polynomial error compounding.

The session drew engaging discussion on reinforcement learning theory, continuous control, plasma stability, partial observability, and the broader role of theoretically grounded AI in complex physical systems.

Learn more here 👇
https://datascience.hku.hk/s20260605fb

衷心感謝來自多倫多大學的牟文龍教授，於 2026 年 6 月 5 日在港大 Inno Wing II 帶來一場精彩的 HKU IDS 嘉賓講座。

是次講座探討機器學習如何回應實用核聚變發展中的核心挑戰之一：維持磁約束等離子體的穩定性。牟教授介紹了用於等離子體控制的可證明模仿學習最新研究，從針對完全觀測模型設計的專家控制器出發，進一步發展可僅依靠實驗中可取得測量數據運作的回饋策略學習算法。

講座亦深入分析線下及線上模仿學習在適應性與穩定性之間的取捨。線下行為複製能夠適應複雜的初始分佈，但可能面對指數級錯誤累積；相比之下，線上算法則為以多項式級錯誤累積實現長期穩定性提供了可行方向。

活動交流涵蓋強化學習理論、連續控制、等離子體穩定性、部分可觀測性，以及具理論基礎的人工智能在複雜物理系統中的應用前景。

講座詳情 👇
https://datascience.hku.hk/s20260605fb

27/05/2026

🎁 Take Home a Piece of HKU IDS
帶走一份 HKU IDS 小心意 🫰

HKU IDS souvenirs are now available at the HKU Visitor Centre, bringing the IDS identity beyond research, programmes, and events into everyday campus life!

Designed for students, alumni, visitors, friends of HKU, and anyone with a passion for data science, the first collection features a range of small, useful, and unmistakably IDS items.

Whether you are picking something for yourself or looking for a small gift, visit the HKU Visitor Centre or shop online to explore the collection!

🛍️ HKU Visitor Centre: G/F, Centennial Campus, HKU
🔗 Shop online: https://datascience.hku.hk/souvenirs-sns (link in bio)

HKU IDS 紀念品現已登陸港大訪客中心！

從研究、課程與活動，延伸至日常校園生活，IDS 誠意為港大同學、校友、訪客、友好，以及醉心於數據科學的每一位，呈獻一系列精緻、實用、充滿 IDS 特色的小物。

無論是自用或饋贈親友，歡迎你親臨港大訪客中心，或到網上商店選購。

🛍️ 港大訪客中心：香港大學百周年校園地下
🔗 網上選購：https://datascience.hku.hk/souvenirs-sns (link in bio)

27/05/2026

🌟 Event Recap 活動回顧 | IDS Scholar Seminar Series – Predictability of Complex Systems 🧠🔗

A big thank you to Prof Qingpeng ZHANG from IDS for delivering an insightful IDS Scholar Seminar today on 27 May 2026 at Inno Wing II, HKU!

The seminar explored the fundamental limits of predictability in complex systems through a statistical physics framework. Prof ZHANG discussed how machine learning problems, including binary classification and link prediction, can be mapped onto spin glass models to better understand the theoretical boundaries of prediction.

The session drew an engaging Q&A discussion both onsite and on Zoom, covering protocol-conditioned predictability, masking strategies, negative sampling, fair comparison between prediction methods, and possible extensions to more complex AI and biological systems.

Learn more here 👇
https://datascience.hku.hk/seminar20260527

衷心感謝 IDS 學者 張清鵬教授，今天（2026 年 5 月 27 日）於港大 Inno Wing II 帶來一場精彩的講座！

是次講座從統計物理框架出發，探討複雜系統中可預測性的理論極限。張教授分享如何將二元分類及鏈接預測等機器學習問題，映射至自旋玻璃模型，以理解預測能力的理論邊界。

活動現場及 Zoom 參與者均參與了問答環節。討論內容涵蓋協議條件下的可預測性、遮蔽策略、負樣本抽取、預測方法之間的公平比較，以及相關框架延伸至更複雜人工智能及生物系統的可能性。

講座詳情 👇
https://datascience.hku.hk/seminar20260527

22/05/2026

🌟 Event Recap 活動回顧 | HKU IDS Guest Seminar – Pointwise Generalization in Deep Neural Networks 🧠

A big thank you to Prof Yunbei XU from the National University of Singapore for delivering an insightful HKU IDS Guest Seminar on 22 May 2026 at Inno Wing II, HKU.

The seminar explored a central question in deep learning theory: why highly complex and over-parameterized neural networks can still generalize effectively. Prof XU introduced a pointwise generalisation framework, offering a spectrum-aware perspective on learned feature representations and their connection with generalization bounds.

The session drew an engaging Q&A discussion on local and global complexity measures, Bayesian and information-theoretic perspectives, optimisation trajectories, and future directions of this research area.

Learn more here 👇
https://datascience.hku.hk/seminar20260522

衷心感謝來自新加坡國立大學的許雲貝教授，於2026年5月22日在港大 Inno Wing II 帶來一場精彩的 HKU IDS 嘉賓講座。

是次講座探討深度學習理論中的核心問題：為何高度複雜、過度參數化的神經網絡，仍能夠有效泛化至未見數據。許教授介紹了一套逐點泛化理論框架，從學習所得特徵表徵的譜結構出發，理解其與泛化界限之間的關係。

活動問答環節交流熱烈，涵蓋局部與全局複雜度量、貝葉斯及資訊理論視角、優化軌跡，以及相關研究的未來發展方向。

講座詳情 👇
https://datascience.hku.hk/seminar20260522