📚 2023 Sloan Research Fellowships: Tsinghua Alumni Han Song and Li Yuanzhi Among the Winners | "诺奖风向标"2023斯隆研究奖出炉：韩松、李远志等清华校友获奖
The Sloan Research Fellowship is one of the most closely watched early-career awards in North America. Its annual list of winners is widely treated as an early indicator of who may become the academic leaders of the next two or three decades, and the 2023 cohort attracted particular attention in Chinese-language academic circles because four of the twenty-two computer science fellows were China-trained scholars, all of them graduates of Tsinghua University.
斯隆研究奖是北美地区面向年轻研究者的一项重要奖项，其获奖者名单常常被视作未来学术领袖的早期风向标。2023 年的获奖名单同样引发关注：在计算机科学领域的 22 位获奖者中，有 4 位华人学者入选，且 4 人皆毕业于中国清华大学。本文将从奖项起源、资助方式、统计意义上的"诺奖风向标"效应、学科覆盖、评选机制，以及本届华人获奖者的代表性研究等角度，系统梳理这一奖项为何值得教育界与科研界持续关注。
1. Why the Sloan Research Fellowship Matters | 斯隆研究奖为何值得关注
Every year the Alfred P. Sloan Foundation announces a new cohort of Sloan Research Fellows, and the announcement is read far beyond the foundation's own offices. University departments, funding agencies, science journalists and graduate applicants all use the list as a rough map of where the most promising young researchers are working right now. Because fellows are selected at a very early stage of their independent careers, the list often names people long before they become famous.
每年，艾尔弗·斯隆基金会都会公布一批新的斯隆研究奖获奖者，而这份名单的影响力远远超出基金会自身。大学院系、资助机构、科学媒体乃至正在申请博士与教职的年轻人，都会把这份名单当作一张"当下最值得关注的年轻研究者分布图"。由于获奖者是在独立科研生涯的极早阶段被选中的，名单上的人往往在成名之前就已经出现。
For students and teachers in international education, the award is also a useful case study. It shows what a research system chooses to reward: not past glory, not institutional brand alone, but the early evidence of originality, independence and future influence. That is a set of criteria worth discussing in any classroom that prepares students for university-level research.
对国际教育领域的学生与教师而言，这一奖项同时也是一个很好的案例：它展示了一个科研体系究竟选择奖励什么——不是过去的荣誉，也不只是机构品牌，而是原创性、独立性与未来影响力的早期证据。对于任何为学生进入大学研究阶段做准备的教学场景来说，这套标准都值得讨论。
2. Origins and Mission of the Award | 奖项起源与定位
The Alfred P. Sloan Foundation established the Sloan Research Fellowship in 1955. The programme was designed specifically for researchers in the United States and Canada who are in the early stages of their careers. Its purpose is to identify and then support scientists who have already demonstrated unusual creativity and innovative potential, rather than to crown a lifetime of achievement.
艾尔弗·斯隆基金会于 1955 年设立了斯隆研究奖。这个奖项专门针对美国以及加拿大地区处于职业生涯早期的研究人员，目的在于识别并支持那些已经展现出非凡创造力与创新潜力的科学工作者，而不是奖励已经完成的终身成就。
That distinction matters. Awards that honour a completed career look backwards; a fellowship of this kind looks forwards. The foundation's core logic is that a relatively small amount of money, given at exactly the right moment, can change the trajectory of a young scientist's work by freeing time, buying equipment and enabling risky but potentially transformative ideas.
这一区别非常关键。表彰终身成就的奖项面向过去，而这类研究奖面向未来。基金会的核心逻辑是：在恰当的时机提供一笔数额并不夸张的资金，能够通过释放时间、购置设备、支持高风险但可能具有变革性的想法，从而改变一位年轻科学家的工作轨迹。
3. The Two-Year Grant and Its Flexible Use | 两年经费与灵活使用方式
Each fellow receives a research grant of 75,000 US dollars, paid over a period of two years. The rules governing the money are deliberately permissive: recipients may spend it on hiring assistants, buying equipment, covering travel to conferences or collaborators, or on anything else that directly advances their research.
获奖者会得到一项研究经费，该经费为期两年，总额为七万五千美元。这笔资金的使用规定相当宽松：研究人员可以自行决定把它用于雇佣助手、购买设备、支付外出差旅，或者用在任何能够直接推动研究进程的方面。
$75,000 ÷ 2 years → a flexible, researcher-controlled budget
This flexible model is intended to maximise the creativity of the researcher and minimise administrative friction. In practice it means a young faculty member can say yes to an experiment, a collaboration or a piece of hardware without waiting for a large external grant to be approved.
这种灵活的资助模式，目的在于最大程度地释放研究人员的创造力，把行政层面的束缚降到最低。在实际运行中，它意味着一位年轻教师可以在不必等待大型外部项目获批的情况下，对一次实验、一项合作或一台设备说"可以"。
4. The "Nobel Prize Bellwether" in Numbers | "诺奖风向标"的统计依据
The nickname "bellwether of the Nobel Prize" is not a marketing slogan; it rests on a long statistical record. According to the foundation's own tallies, 51 former Sloan Research Fellows have gone on to win a Nobel Prize since 2007 alone, which is a striking demonstration of how well the selection process identifies top scientific talent at an early stage.
该奖项被称作"诺奖风向标"，并非宣传口号，而是有着坚实的统计数据作为支撑。据统计，仅自 2007 年以来，在斯隆奖的历届获得者之中，就已经有 51 人在之后获得了诺贝尔奖，这清楚地展现了该奖项在早期识别顶尖科学人才方面具备的卓越眼光。
Nor is this predictive power limited to physics and chemistry. In the same period, the fellowship's alumni have also collected 17 Fields Medals, 69 United States National Medals of Science and 20 John Bates Clark Medals, the last being the most prestigious honour for young economists in the United States.
这种预测能力并不局限于物理学与化学等自然科学领域。在同一时期中，其获奖者里还出现了 17 位菲尔兹奖得主、69 位美国国家科学奖章获得者，以及 20 位约翰·贝茨·克拉克奖得主——后者是美国授予年轻经济学家的最高荣誉。
| Later honour | 后续荣誉 | Number of fellows | 获奖者人数 |
| Nobel Prize | 诺贝尔奖 | 51 (since 2007 | 2007 年以来) |
| Fields Medal | 菲尔兹奖 | 17 |
| US National Medal of Science | 美国国家科学奖章 | 69 |
| John Bates Clark Medal | 克拉克奖 | 20 |
Taken together, these figures confirm the success of the selection committees in spotting future academic leaders rather than simply rewarding fashionable topics.
这些数字共同证实了评审委员会在挑选未来学术领袖方面的成功，也说明这一奖项并非只是在追捧一时的热门方向。
5. Seven Disciplines Currently Covered | 当前涵盖的七个学科领域
The Sloan Research Fellowship currently covers seven scientific and technical fields: chemistry, computer science, Earth system science, economics, mathematics, neuroscience and physics. Each year roughly twenty fellows are selected in each field, giving a cohort of around 125 researchers in total.
当下，斯隆研究奖的评选范畴涵盖七个科学与技术领域，分别是化学、计算机科学、地球系统科学、经济学、数学、神经科学以及物理学。每年各领域平均甄选出大约二十位获奖者，因此一届的总规模通常在 125 人左右。
This distribution reflects the foundation's support for basic science, frontier technology and the crossings between them. Earth system science, for example, sits at the intersection of physics, chemistry, biology and data modelling, while computer science increasingly reaches into mathematics, neuroscience and physics.
这样的领域设置体现出基金会对基础科学、前沿技术以及二者交叉融合的支持导向。例如地球系统科学处于物理学、化学、生物学与数据建模的交汇处，而计算机科学则越来越多地延伸到数学、神经科学与物理学之中。
- Chemistry | 化学
- Computer Science | 计算机科学
- Earth System Science | 地球系统科学
- Economics | 经济学
- Mathematics | 数学
- Neuroscience | 神经科学
- Physics | 物理学
6. How Fellows Are Selected Each Year | 每年评选机制
Selection is carried out by committees of senior scholars in each discipline, and candidates are normally nominated rather than simply applying. The evaluation stresses the originality and independence of the candidate's research, and its potential influence on the field over the years ahead. There is no restriction on the specific research direction a candidate chooses.
评选进程由各学科领域的资深学者所构成的委员会加以严格评审，候选人通常需要被提名，而不是单纯自行申请。评审着重考量候选人研究的原创性、独立性，以及其未来对学科可能产生的影响力，并且不对候选人的具体研究方向设置限制。
The emphasis on independence explains why the fellowship is often described as a bet on a person rather than on a project. Reviewers are asked, in effect, whether this researcher has already shown the ability to set an agenda and carry it forward without standing in a supervisor's shadow.
对独立性的强调，解释了为什么这一奖项常被形容为"押注于人"而不是"押注于项目"。评审人实际上要回答的问题是：这位研究者是否已经展现出自己设定研究议程、并独立推进它的能力，而不是始终站在导师的影子之下。
7. Chinese Scholars in Computer Science in 2023 | 计算机科学领域的华人学者表现
In the 2023 round, computer science produced 22 fellows, four of whom were Chinese scholars. All four completed their undergraduate or doctoral training at Tsinghua University: Han Song, Yuanzhi Li, Chenhao Tan and Xiaodi Wu, who now hold faculty positions at leading North American universities.
本年度计算机科学领域的获奖者有 22 位，其中有 4 位华人学者入选。这 4 位华人学者均毕业于中国清华大学，分别是韩松、李远志、谭宸浩与伍骁迪，他们目前都在北美一流大学担任教职。
Their research directions span artificial intelligence model compression, the theoretical foundations of deep learning, natural language processing and machine learning systems, and quantum computing. Together they illustrate both the international competitiveness of Chinese computer science training and the growing presence of scholars of Chinese origin on the global computing research stage.
他们的研究方向涵盖人工智能模型压缩、深度学习理论基础、自然语言处理与机器学习系统，以及量子计算等一个又一个前沿热点。这一现象既展现出清华大学在计算机科学人才培养方面的国际竞争力，也反映出华裔学者在全球计算研究舞台上愈发强大的活跃度与不断提升的影响力。
8. The Four Alumni and Their Institutions | 四位校友与他们的任职机构
Han Song is an associate professor at the Massachusetts Institute of Technology, where he works on efficient machine learning and computer systems. Yuanzhi Li is an assistant professor at Carnegie Mellon University, with a focus on the theory of deep learning. Chenhao Tan is an assistant professor at the University of Chicago, working at the intersection of natural language processing and human-centred artificial intelligence. Xiaodi Wu is an assistant professor at the University of Maryland, working on quantum information and computation.
韩松是麻省理工学院副教授，研究方向包括高效机器学习与计算机系统。李远志是卡内基·梅隆大学助理教授，主要关注深度学习理论。谭宸浩是芝加哥大学助理教授，研究自然语言处理与以人为本的人工智能的交叉方向。伍骁迪是马里兰大学助理教授，从事量子信息与计算研究。
What links these four careers is a shared emphasis on making computation more powerful, more efficient or more theoretically grounded. That combination is precisely what the fellowship's reviewers tend to reward: work that is ambitious in scope yet still grounded in concrete technical problems.
把这几条学术路径联系在一起的，是一种共同的取向：让计算变得更强、更高效，或者拥有更坚实的理论基础。这恰恰也是该奖项评审者倾向于奖励的组合——研究视野足够宏大，同时又牢牢落在具体的技术问题上。
9. Representative Research: Han Song | 获奖学者代表性研究：韩松
Han Song's best-known contribution is deep model compression, work that was recognised with the ICLR 2016 Best Paper Award. The idea is to take a large neural network and reduce its size and computational cost dramatically while keeping most of its accuracy.
以韩松而言，他具有代表性的工作涵盖深度模型压缩技术，此项成果曾荣获 ICLR 2016 最佳论文奖。其核心思路是：在尽量保持精度的前提下，大幅压缩大型神经网络的体积与计算开销。
He also developed the ESE sparse neural network inference engine, which received the FPGA 2017 Best Paper Award. Such systems aim to make complex artificial intelligence models run effectively on devices with limited computing resources, from embedded hardware to mobile phones.
他所开发的 ESE 稀疏神经网络推理引擎同样获得了 FPGA 2017 最佳论文奖。这类系统致力于让复杂的人工智能模型能够在计算资源受限的设备上有效运行，从嵌入式硬件到移动电话都属于其应用场景。
Large model → compression → sparse inference → deployment on constrained hardware
The wider significance is that efficiency research translates directly into accessibility. A model that once required a data centre can, after compression, run on a device that a student can hold in one hand.
更广泛的意义在于，效率研究可以直接转化为可获得性。原本需要数据中心才能运行的模型，经过压缩后，可以在一台学生单手就能握住的设备上运行。
10. Representative Research: Yuanzhi Li and Xiaodi Wu | 李远志与伍骁迪的研究
Yuanzhi Li's work focuses on the theoretical foundations of deep learning, including the design and analysis of optimisation algorithms. Understanding why gradient-based training works, and when it fails, is one of the central open questions of modern machine learning.
李远志所进行的研究，着重于深度学习的基础理论方面，其中包含优化算法的设计以及分析。理解基于梯度的训练为何有效、何时失效，是现代机器学习最核心的开放问题之一。
Xiaodi Wu works in the theory of quantum information and computation, with research covering quantum cryptography and quantum communication. His field asks how the laws of physics themselves can be used as a computational resource, and how information can be protected by those laws rather than by computational assumptions alone.
伍骁迪则在量子信息与计算的理论范畴展开工作，其研究涉及量子密码学以及量子通信。这一领域探讨的问题是：物理定律本身能否被当作一种计算资源，以及信息能否由物理定律本身、而非仅靠计算复杂性假设来加以保护。
Taken together, these four careers span the full arc of computing, from software algorithms at one end, through hardware and systems in the middle, to physical principles at the other. That arc is exactly what the phrase "frontier of computer science" is meant to describe.
把这几条路径合起来看，它们涵盖了计算科学的完整链条：一端是软件算法，中间是硬件与系统，另一端则触及物理原理。这正是"计算机科学前沿"这一说法所描述的图景。
11. What This Means for Research Education | 对科研教育体系的启示
The selection criteria of the Sloan Research Fellowship offer a useful reference point for higher education and for the training of research talent. The award emphasises the independent innovative capacity of a researcher in the early career stage rather than relying on the reflected prestige of a supervisor or a platform, which provides a model for how young researchers can be evaluated and supported.
斯隆奖的评选标准对高等教育以及科研人才培养具有参考价值。它突出研究者职业生涯早期阶段所拥有的独立创新能力，而不是单纯依靠导师或平台所散发的光环，这为如何评估与资助青年科研人员提供了范本。
The inclusiveness of the award with respect to research direction, and the flexibility with which its funds may be used, are equally worth borrowing. Such a model encourages researchers to explore high-risk, highly innovative topics, and that kind of support is helpful for incubating breakthrough results.
奖项在研究方向上的包容性，以及资金使用上的灵活性，同样值得借鉴。它能够激励研究者大胆探索高风险、高创新性的课题，而这种支持模式有助于孕育突破性的科研成果。
- Early-career independence matters more than pedigree | 早期独立性比出身标签更重要
- Small, flexible funding can unlock large ideas | 小额而灵活的资金可以撬动大想法
- No topic restriction keeps interdisciplinary work alive | 不限定题目方向有助于跨学科研究存活
- Long-term tracking creates credible statistics | 长期追踪才形成可信的统计证据
For schools and universities, the practical lesson is that supporting young researchers is not only about money. Time, freedom to choose a direction, tolerance of failure and access to a genuine research community are often just as decisive.
对于学校与大学而言，其实践启示是：支持青年科研人员不仅仅是给钱的问题。时间、选择方向的自由、对失败的容忍度，以及进入一个真正的研究共同体的机会，往往同样具有决定性。
12. Key Takeaways and a Question for Readers | 要点总结与读者提问
To summarise: the Sloan Research Fellowship was founded in 1955 for early-career researchers in the United States and Canada; it provides 75,000 US dollars over two years with unusually flexible spending rules; it covers seven fields and roughly twenty fellows per field each year; and its alumni include 51 Nobel laureates since 2007, along with 17 Fields Medalists, 69 National Medal of Science recipients and 20 John Bates Clark Medalists.
概括而言：斯隆研究奖于 1955 年设立，面向美国与加拿大的早期科研人员；提供为期两年、总额七万五千美元、使用规则异常灵活的经费；涵盖七个学科领域，每年每个领域约二十人；其获奖者中自 2007 年以来走出 51 位诺贝尔奖得主，另有 17 位菲尔兹奖得主、69 位美国国家科学奖章获得者与 20 位克拉克奖得主。
In 2023, four of the twenty-two computer science fellows were Chinese scholars trained at Tsinghua University, working on topics ranging from model compression and deep learning theory to quantum information and human-centred AI. Their recognition is both a personal milestone and a signal about the international competitiveness of the training they received.
2023 年，计算机科学领域 22 位获奖者中有 4 位是毕业于清华大学的华人学者，研究方向从模型压缩、深度学习理论延伸到量子信息与以人为本的人工智能。他们的获奖既是个人里程碑，也是其所受培养具有国际竞争力的一个信号。
For you, the reader: beyond providing funding, what other factors are most effective in activating the innovative potential of young researchers? You are welcome to share your views in the comments, and if this article has been useful to you, a like is always appreciated.
那么，对您而言，在助力青年科研人员方面，除了提供资金之外，还有哪些关键要素能够以最有效的方式激活他们的创新潜能？欢迎在评论区分享您的看法。如果这篇文章对您有所启发，也欢迎点赞支持。
Published by TutorHao | Science News | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图