报 告 人：邓东灵，清华大学交叉信息研究院特别研究员
报告地点：电子楼207会议室
报告时间：2025年3月12日（周三）上午9:30
报告题目：Quantum automated learning with provable and explainable trainability
个人简介:
邓东灵，清华大学交叉信息研究院特别研究员，博士生导师，海外高层次人才青年项目、国家杰出青年科学基金获得者。2007年获南开大学物理、数学双学士学位，2015年博士毕业于美国密西根大学，博士论文获“Kent M. Terwilliger Memorial Thesis Prize”奖。2015-2018年在马里兰大学联合量子研究所从事博士后研究，2018年回国入职清华大学。现任剑桥大学出版社杂志《Research Directions: Quantum Technologies》执行编辑，《njp Quantum Information》副编辑，《Communications in Theoretical Physics》、《Quantum Review Letters》编委。2021年获天津市自然科学一等奖（第二完成人），2022年获年度清华大学“先进工作者”称号。主要研究方向为量子人工智能，在Nature, Nature/Science子刊，PRL/PRX等期刊上发表论文百余篇。
报告简介：Quantum artificial intelligence, an emerging research frontier at the intersection of quantum computing and machine learning, holds unprecedented potential to revolutionize AI technologies. However, the field remains in its infancy, with fundamental challenges impeding its scalability and real-life applications. In this talk, I will first give a brief introduction to quantum AI and the challenges faced in scaling up. Then, I introduce a new paradigm, called quantum automated learning, that circumvents these challenges. This approach is inherently gradient-free and scalable, with provable and explainable trainability. I will give a couple of concrete examples regarding classification of both real-life images and quantum datasets to show its effectiveness and sample efficiency.
报告人：李炜康，清华大学博士研究生
报告地点：电子楼207会议室
报告时间：2025年3月12日（周三）上午9:30
报告题目：Quantum Optimization and Machine Learning——from Fundamental Physics to Practical Applications
个人简介:
李炜康，清华大学交叉信息研究院博士研究生。2020年本科毕业于中国科学技术大学少年班学院，获应用物理学学位。主要研究兴趣包括量子机器学习理论，量子-经典优化算法，机器学习在量子物理中的应用，量子信息理论等。在Nat. Comput. Sci.，Nat. Comm.，Rep. Prog. Phys.等期刊发表了量子人工智能领域的工作，并在超导量子平台上首次合作演示了量子对抗学习。当前的研究方向包括研究量子学习优势、利用量子资源设计在计算复杂性和隐私保护方面具有可证明优势的量子学习框架；以及利用人工智能算法辅助量子物理研究、量子线路编译等。
报告简介：The intersection of quantum science and AI holds transformative potential for tackling some of the most challenging problems in both fields. Quantum algorithms, particularly in quantum optimization and quantum machine learning, offer powerful solutions to problems that surpass the limits of classical computational methods. Conversely, AI techniques also provide innovative tools to address complex questions in quantum physics. In this talk, I will mainly focus on two key applications of quantum optimization and quantum machine learning: (1) detecting Bell-operator correlations and (2) quantum delegated learning, highlighting their implications for both fundamental physics and practical tasks. Additionally, I will explore how classical optimization and AI algorithms contribute to advancing quantum physics, such as by establishing optimizable mappings between Bell inequalities and Hamiltonians, and enabling efficient quantum circuit compilation for many-body experiments. Finally, I will outline future research directions in this exciting field, emphasizing the mutual enrichment of quantum science and AI.
