微软亚洲研究院铸星计划

星亮点

加入2024微软亚洲研究院铸星计划，您将迎来星辉璀璨的科研之旅：

星搭档：加入世界一流的企业研究院，与MSRA资深研究员携手探索新边界。
星项目：有机会参与微软内部科研项目，获得微软计算、工具和数据的支持。
星平台：获得微软国际化学术平台的支持，开展国际与多边合作。
星资助：优先获得微软亚洲研究院科研合作项目资助，开启长期合作。
星友会：加入铸星计划星友会，与全球铸星学者建立联系，共同成长。

星轨迹

2023年11月1日：申请启动
2023年12月31日：申请截止
2024年1月1日-2月19日：访问方案确认和选拔
2024年2月20日：结果公布
2024年2月21日-2025年1月31日：2024年度访问开始

星申请

申请人来自微软亚洲研究院合作高校，且获得相关领域专家和所在院系单位推荐。
申请人需为大学正式教职人员且满足博士毕业不超过7年（包括7年）的年限要求。
申请人的研究领域应在铸星计划项目关注的研究领域之内（详见“铸星领域”页面）。
项目要求为期三个月的全职访问；有特殊情况者，在获得项目委员会批准的前提下，在三个月的现场访问中最多可申请到一个月远程访问。

星报名

点击注册链接，填写报名信息并提交，以明确报名意向。
下载项目申请表(中文)、专家推荐信(中文) 和单位推荐信(中文) 模板，根据模板信息提供正式的申请信息，并邮件至微软亚洲研究院郑梦真，邮箱：v-menzheng@microsoft.com。

如有疑问，请邮件咨询微软亚洲研究院铸星计划项目负责人石贝贝 besh@microsoft.com。

1. Societal AI

Big models have showcased unprecedented achievements, establishing a new frontier in technological advancements. Nevertheless, these models are accompanied by profound societal challenges, encompassing security vulnerabilities, ethical considerations, intricate evaluation mechanisms, and broader social repercussions. This research theme aims to foster a comprehensive and interdisciplinary examination of these issues, juxtaposing AI dynamics with parallels in human development.

Value Alignment: An exploration into ensuring that AI systems resonate with the ethical and moral principles we aim to impart. The goal is to understand and address the discrepancies between human values and AI outcomes.
LLM Safety: Delving into the protection of LLMs against malicious interventions and potential misuses. The research will also focus on developing defensive strategies and establishing safety best practices for these models.
LLM Evaluating: A study dedicated to devising robust and holistic performance assessment techniques for AI systems. This will include identifying suitable metrics and benchmarks that capture the multifaceted nature of these models.
Interdisciplinary Collaborations: By bridging AI with social sciences, we aim to deeply comprehend and tackle the societal ramifications of AI. This involves dissecting its impact on individual, group, and societal dynamics.

Xing Xie (Engaging Lead), Senior Principal Research Manager, Microsoft Research Asia

With the rising prominence of AI and its profound influence on society, I truly hope that your visit will ignite innovative perspectives and enhance our approach to AI. Additionally, we're keen to explore the integration of AI with social sciences, and your insights are invaluable in bridging this divide. As we advance technologically, it remains crucial to maintain a holistic view. I look forward to the collaborative strides we will make together.

Fangzhao Wu, Principal Researcher, Microsoft Research Asia

We hope the applicants have real passion in studying and improving the safety of LLMs against many serious risks such as Jailbreaking and Prompt Injection Attack. We also welcome applicants who want to study the LLM’s impact on society, such as productivity, employment, education, and research.

Jianxun Lian, Senior Researcher, Microsoft Research Asia

To obtain a comprehensive understanding of the societal challenges and impacts of LLMs, it is insufficient to solely focus on the individual LLM's perspective. We advocate for a thorough investigation of AI-related societal topics from diverse viewpoints, such as integrating methods and theories across various disciplines or employing multiple, interconnected LLMs as a foundation. We anticipate a profound exchange of knowledge between artificial intelligence and other disciplines, including social science, psychology, and urban computing and ultimately, yielding interdisciplinary insights and uncovering novel interdisciplinary applications during this research visit.

Jindong Wang, Senior Researcher, Microsoft Research Asia

We hope to gain a holistic understanding of the evaluation of LLMs during this visit. Specifically, establishing a unified and general evaluation framework to support different scenarios and protocols as well as different interdisciplinary research, cultivating new evaluation algorithms and analysis tools to offer insights for assessment, and building a collaborative environment to unite potential researchers in this area.

Xiaoyuan Yi, Senior Researcher, Microsoft Research Asia

We expect to establish a systematic solution to aligning LLMs' intrinsic values with those of humans by incorporating interdisciplinary insights and methodologies during this visit. Our research will delve into topics include investigating the underlying values of LLMs and their corresponding social impact, and then fostering novel algorithms to align LLMs' values in a unified, flexible, and efficient way, paving the way for a future where LLMs consistently serve the greater good.

2. Media Foundation

Media Foundation is a cutting-edge research theme that delves into the convergence of multimedia and advanced AI technologies to revolutionize our lives in ways yet to be imagined. Drawing from various disciplines, Media Foundation aims to investigate the potential of advanced AI to learn from real-world multimodal media data, focusing on the symbiotic evolution of multimedia and AI. Humans possess remarkable learning capabilities, acquiring a wide range of skills and knowledge from various modalities, such as vision, hearing, touch, and language. Advanced AI systems also need to learn from the real world. However, how can we bridge the gap between the complex and noisy environment and the abstract and semantic representations of advanced AI? By transforming physical and virtual worlds into media tokens, media foundation research has the potential to foster AI learning and creativity, paving the way for innovative multimodal applications.

Neural codecs
Neural communication
Computer vision
Audio and speech
Foundation models
Deep learning fundamentals for multimedia

Yan Lu (Engaging Lead), Partner Research Manager, Microsoft Research Asia

We welcome applicants who possess a strong background in multimedia, AI, or related fields, including individuals with experience in neural codecs, computer vision, deep learning, audio and speech, and other related domains. We also expect participants to be passionate about exploring the paradigm shifts in video and audio research and are eager to contribute to the development of advanced AI systems that can enhance human learning and creativity.

Xiulian Peng, Principal Research Manager, Microsoft Research Asia

We welcome applicants with expertise in audio/speech, cross-modal learning, AI, and related fields, especially audio/speech technologies. We expect individuals to be passionate about delving into the paradigm shift within audio/speech-related research and be eager to contribute to the advancement of cutting-edge AI systems that can augment human learning and foster creativity.

Cuiling Lan, Principal Researcher, Microsoft Research Asia

We look forward to collaborating with visiting researchers who have great passion for media foundation research. We warmly welcome applicants who: 1) possess a deep passion for exploring innovative approaches in multimedia and AGI research; 2) have a strong background in AI, multimedia, or a related field, such as computer vision, multi-modal understanding.

Bin Li, Principal Researcher, Microsoft Research Asia

We welcome candidates with expertise in multimedia, AI, and related fields, such as neural codecs, computer vision, and deep learning. We value individuals passionate about investigating video research's paradigm shifts and contributing to advanced AI systems development, fostering human learning and creativity enhancement. We expect applicants to have a deep understanding of the field and are committed to pushing the boundaries of AI and multimedia research.

3. AI for Sustainability

In the face of accelerating climate change, the urgency of achieving sustainability and carbon neutrality has become a global imperative. Microsoft, recognizing the transformative potential of artificial intelligence (AI), introduces the research theme "AI for Sustainability." This initiative aims to harness AI's unparalleled capacity for data analysis, prediction, and system optimization to address pressing environmental challenges. With a focus on four key pillars—carbon emission monitoring, renewable energy generation optimization, battery management and power-grid optimization, and power consumption reduction—we endeavor to create innovative AI solutions that directly contribute to global carbon neutrality goals. Our commitment extends to producing high-impact research papers for prestigious journals and conferences. This initiative not only underscores Microsoft's dedication to leveraging technological advancements for global good but also sets a benchmark for integrating AI into sustainable practices.

AI for carbon emission monitoring
AI for renewable energy generation optimization
AI for battery management
AI for power-grid optimization
AI for power consumption. reduction

Jiang Bian (Engaging Lead), Senior Principal Research Manager, Microsoft Research Asia

I am eager to welcome visiting researchers to Microsoft’s “AI for Sustainability” initiative, a groundbreaking endeavor committed to utilizing artificial intelligence (AI) in addressing crucial environmental challenges and propelling us toward global carbon neutrality. As we navigate through the complexities of accelerating climate change, your expertise and dedication will be invaluable, particularly in our focused areas of carbon emission monitoring, renewable energy optimization, battery and power-grid management, and power consumption reduction. I anticipate a collaborative spirit, innovative thinking, and a steadfast commitment to excellence as we collectively strive to create AI-driven solutions with real-world impact. Your contributions will not only play a pivotal role in advancing our research initiatives but also in achieving tangible results that resonate on a global scale. I am confident that together, we will produce high-impact research, share knowledge, and forge new paths in integrating AI into sustainable practices, setting a new benchmark for technological innovation for the greater good.

Lei Song, Principal Researcher, Microsoft Research Asia

As a passionate advocate for environmental sustainability, I am eager to explore the potential of AI in optimizing renewable energy generation, power-grid management, and power consumption reduction. I am confident that my background in industrial innovations, combined with experience in AI research, will enable us to make meaningful contributions to this collaboration.The present global energy landscape is undergoing a swift shift from fossil fuels to renewable energy sources, introducing challenges in optimizing efficiency, maintaining a reliable power supply, and enhancing grid management. I am convinced that AI holds the potential to transform the energy sector by tackling these obstacles and fostering more sustainable energy systems. Please feel free to get in touch with us if you share our vision and enthusiasm for these subjects.

Shun Zheng, Senior Researcher, Microsoft Research Asia

I expect to collaborate on research topics related to Lithium-ion batteries, electrical vehicles, and renewable energy solutions. The main topic will be applying cutting-edge machine learning techniques into critical research problems in these domains.

Jinyu Wang, Senior Applied Scientist, Microsoft Research Asia

We warmly welcome visiting researchers who have a keen interest in environmental sustainability and possess expertise in AI, renewable energy, or related fields. If you have experience in areas such as optimizing renewable energy generation, power-grid management, power consumption reduction, and AI-driven innovations in the energy sector, we would be delighted to have you join us. Our visiting researchers are encouraged to be enthusiastic about discovering new ways to leverage AI technology in addressing the challenges faced by the global energy landscape and contribute to the development of more efficient and sustainable energy systems. If you share our passion and dedication to transforming the energy sector through AI-based solutions, we wholeheartedly invite you to collaborate with us in making a significant impact on the future of renewable energy.

Xiaofan Gui, Applied Scientist, Microsoft Research Asia

We welcome applicants with a strong background in renewable energy, electric vehicles, and battery manufacturing, as well as individuals experienced in sustainability and climate change research, ultimately contributing to the development of advanced AI systems that address real-world challenges and promote sustainable practices.

以上三个研究领域是铸星计划重点关注的合作方向。此外，申请人还可以从以下领域中选择自己感兴趣的方向: Heterogeneous Extreme Computing, Intelligent Cloud and Edge, Intelligent Multimedia, Internet Graphics, Machine Learning, Media Computing, Multi-Modal Computing, Natural Language Computing, Networking Infrastructure, Social Computing, Systems, Trustworthy Systems, Visual Computing, Wireless.

项目流程

项目申请与前期沟通：符合项目申请条件的申请人，请按照项目报名要求准备申请表、专家推荐信以及单位推荐表，并于2023年12月31日项目报名截止前通过注册链接填写报名信息并将正式申请表和推荐信邮件发送至微软亚洲研究院学术合作部郑梦真，邮箱：v-menzheng@microsoft.com。在此期间，请申请人与访问期间想要开展合作的微软亚洲研究院研究员建立联系并就访问项目和安排意向事宜做前期沟通（如需帮助，请联系项目负责人石贝贝，邮箱：besh@microsoft.com）。
项目对接与评审：2024年1月，微软亚洲研究院铸星计划项目委员会将与合作研究员们对全部申请材料进行评审，选择和确定合适的候选人制定访问计划，并公布最终名单。
项目邀约：微软亚洲研究院铸星计划项目委员会将以书面形式向最终名单上的青年学者发送“微软亚洲研究院铸星计划邀请函”，预计发件时间为2024年2月20日。
项目访问计划确定：申请人在收到“微软亚洲研究院铸星计划邀请函”后，请与项目负责人石贝贝以及相关合作研究员一同确定具体访问方案。铸星计划2024访问周期自2024年2月开始到2025年1月结束。请注意，申请人由于个人原因自收到书面邀请函之日起一年未能达成访问计划的，将自动被视为放弃参与该年度铸星计划。
项目访问计划变更：微软亚洲研究院铸星计划项目要求为期三个月的全职现场访问；可在项目负责人的帮助下，向项目委员会提出访问计划变更的申请。包括以下两种情况

- 无法满足现场访问三个月。请在访问开始之前于访问计划确定时联系项目负责人以及相关合作研究员商议并提出访问计划变更申请，获得项目委员会批准后，在三个月的现场访问中最多可申请到一个月远程访问；访问开始后不可进行此项变更申请。
- 申请三个月以上访问时间。申请人在访问期间可根据项目需要，联系项目负责人以及相关合作研究员商议并提出延长访问时间的申请，获得项目委员会的批准后，最多可再延长3个月现场访问（不可申请延长远程访问）；此项申请最晚需在访问结束前30天提出方可有效。

项目访问结束：参与铸星计划的铸星学者在项目访问结束时，需按要求完成书面的访问项目结题报告，并在微软亚洲研究院内部进行一场学术演讲。项目访问结束后，铸星学者将受邀加入微软亚洲研究院铸星计划星友会。