2023年卓越領域-數量歷史研討會:探索中國和香港的資料與研究之旅

The Centre for Quantitative History of HKU Business School hosted a two-day conference on October 25-26, 2023, showcasing the latest quantitative history (QH) research carried out under the Areas of Excellence (AoE) Scheme funded by the Research Grants Council. This scheme aims to increase research collaboration. The conference brought together over 60 attendees from Asia, Europe, and North America to explore the latest trends and ideas in quantitative history on the HKU campus. It focused on four theme clusters that examined key dimensions of how China and Hong Kong reached their current positions: 1) ancient roots, 2) culture, 3) state capacity & institutions, and 4) finance, trade & Hong Kong.

The conference featured 16 ongoing research projects from the project team members and their collaborators. These projects applied the quantitative history approach, utilizing large datasets constructed from historical archives and archaeological records to reconstruct trends, and patterns, and assess causal relationships.

The conference opened with the first theme, “Ancient Roots from Quantitative Perspectives.” During the presentation on war and technology, Professor Zhiwu Chen, the Director of the Centre for Quantitative History and the Project Coordinator of the AoE project, highlighted that China offered a longstanding puzzle to historians in that its society had produced an astonishing number of inventions – including printing, the blast furnace, and gunpowder – but had never exploited them to their full potential. The second theme, “Culture, Religion, and Long-Term Consequences,” explored natural and human-created disasters using meticulously collected data, for example, the China Government Employee Database – Qing Jinshenlu (CGED-Q JSL) datasets collected by Co-PI Professor Cameron Campbell and his team of HKUST.

On the second day of the conference, the focus shifted to the theme “State Capacity, Institutions, and Development,” which was highlighted by five paper presentations, further emphasizing its popularity in the research landscape. Notably, Co-PI Professor Debin Ma of Oxford presented his 15-year research on a Millennium of Public Finance. The final theme, “Financial History, Trade, and the Rise of Hong Kong as a Financial Centre,” highlighted cutting-edge research in the field of quantitative history led by Co-PI Professor Chicheng Ma of HKU Business School. It is worth mentioning that all three papers in this cluster were either presented or co-authored by postgraduate research students, demonstrating the project team’s commitment to nurturing young scholars in this interdisciplinary research landscape. The conference also featured six research student presenters from HKU, HKUST, NUS, and Boston University.

The conference organizing committee was pleased to have four discussants join the team’s efforts to discuss the presented work. These included CQH Advisory Board member Professor Tetsuji Okazaki of the University of Tokyo, Professor Paul Seabright of the University of Toulouse, Professor Tuan-Hwee Sng of the National University of Singapore, and Professor Jin Li of HKU Business School. Their generous contributions and discussions were greatly appreciated by the team members and co-authors.

In the closing remarks, Professor Chicheng Ma, also the Associate Director of the Centre for Quantitative History, expressed gratitude to all the participating researchers for their support in strengthening interaction and collaboration in this ever-growing field. Co-PI Professor Cameron Campbell, Associate Dean of Humanities and Social Science at HKUST and the AoE-QH Team leader in external relations expressed delight, along with the team members, in the conference’s atmosphere, engagement, and presentations, ranking it among the top-tier workshops they have attended.

Stay tuned for upcoming AoE-QH events and activities in Hong Kong!

The two-day Area of Excellence-Quantitative History Conference 2023 gathered project team members (listed in alphabetical order): Zhiwu Chen (Project Coordinator), Jed O. Kaplan, Chen Lin, and Chicheng Ma from HKU; Ying Bai (CUHK); Cameron Campbell (HKUST); Zhan Lin (Renmin); William Guanglin Liu (Lingnan); and Debin Ma (Oxford), along with their collaborators, co-authors, invited discussants, and guests on the HKU campus.

Highlighting young scholar contributions: Three conference papers co-authored by postgraduates, with six student presenters from top universities.

Group Photo: Day 1 of the conference outside the historic May Hall

Group Photo: Taken before the conference participants headed off for the closing dinner.

 

Hi-res photos are available here “Day 1” and “Day 2”.

Other Events
港大經管學院最新多模態AI圖像生成能力排名出爐 部份中國人工智能模型表現突出
2025 | 學院新聞
港大經管學院最新多模態AI圖像生成能力排名出爐 部份中國人工智能模型表現突出
港大經管學院今日發表《人工智能模型圖像生成能力綜合評測報告》,針對15個「文生圖模型」及7個「多模態大語言模型」進行全面評估。研究顯示,字節跳動的即夢AI和豆包,以及百度的文心一言,在新圖像生成的內容質素及圖像修改的表現突出;而早前引起全球關注的DeepSeek最新推出的文生圖模型Janus-Pro,則在新圖像生成方面表現欠佳。研究亦發現部分文生圖模型雖然在內容質素方面表現優異,卻在安全與責任方面的表現強差人意。整體而言,與文生圖模型相比,多模態大語言模型整體表現較佳。
人工智能圖像生成評測成績單公布: 字節跳動百度表現亮眼,DeepSeek Janus-Pro表現欠佳
2025 | 研究
人工智能圖像生成評測成績單公布: 字節跳動百度表現亮眼,DeepSeek Janus-Pro表現欠佳
如今,人工智能領域的前沿模型技術已經從文本處理拓展至視覺信息的深度理解與生成。這些模型既能精准解讀圖像語義,又能根據文字描述創作出兼具真實感與藝術性的視覺內容,展現出令人驚嘆的跨模態理解與創作能力。本研究聚焦全新圖像的生成和基于現有圖像的圖像修改兩大核心任務,提出了一套系統性的人工智能模型圖像生成能力評測框架。我們基于多維測試集的構建與專家評審,對15個專業文生圖模型和7個多模態大語言模型的圖像生成能力進行了全面評估。結果顯示,字節跳動的即夢AI和豆包以及百度的文心一言在新圖像生成的內容質量與修改任務中表現突出,位列第一梯隊。對比不同類型的AI模型,我們發現,相對于專業文生圖模型,多模態大語言模型整體表現更佳。