Unverified Commit 1d2c9080 authored by liziwl's avatar liziwl Committed by GitHub
Browse files

Merge pull request #191 from SUSTech-CRA/poster-update-2025-04-05T11-46-57

Add docs/study/talks
parents e1f7647c f2861eb6
Loading
Loading
Loading
Loading
+15 −0
Original line number Diff line number Diff line
# 李竹:多尺度稀疏卷积学习在点云压缩、超分辨率和去块效应中的应用

* 题目: 多尺度稀疏卷积学习在点云压缩、超分辨率和去块效应中的应用
* 主讲人:李竹 @ 美国密苏里大学堪萨斯分校教授
* 时间:2025年3月14日 10:30 - 12:00
* 地点:工学院南楼326会议室

## 主讲人简介
Zhu Li is now a professor with the Dept of Computer Science & Electrical Engineering (CSEE), University of Missouri,Kansas City, and the director of the NSF Center for Big Learning at UMKC. He was Sr. Staff Researcher/Sr. Manager with with Samsung Research America's Multimedia Standards Research Lab in Richardson, TX, 2012-2015, Sr. Staff Researcher/Media Analytics Group Lead with FutureWei (Huawei) Technology's Media Lab in Bridgewater, NJ, 2010~2012, and an Assistant Professor with the Dept of Computing, The Hong Kong Polytechnic University from 2008 to 2010, and a Principal Staff Research Engineer with the Multimedia Research Lab (MRL), Motorola Labs, from 2000 to 2008.

## 讲座简介
Due to the increased popularity of augmented and virtual reality experiences, as well as 3D sensing for auto-driving, the interest in capturing high resolution real-world point clouds has grown significantly in recent years. Point cloud is a new class of signal that is non-uniform and sparse and this present unique challenges to the signal processing, compression and learning problems. In this talk, we present our multi-scale sparse convolutional learning and Graph Fourier Transform (GFT) based framework for large scale point cloud processing, with applications to the geometry and attributes super-resolution, and dynamic point cloud compression with latent space compensation. The architecture is memory efficient and can learn deep networks to handle large scale point cloud in real world applications. Initial results demonstrated that this framework achieved new state of the art results in geometry super-resolution, attributes deblocking and super-resolving, and dynamic point cloud sequence compression.

## 海报链接
![](https://gtimg.liziwl.cn/post-img/2025-03-14T10-30-00_%E6%9D%8E%E7%AB%B9.jpg)
 No newline at end of file
+15 −0
Original line number Diff line number Diff line
# 《科学大讲堂第188期》Prof. Peter May:有限集、对称群与范畴组合

* 题目: 有限集、对称群与范畴组合
* 主讲人:Prof. Peter May @ 美国芝加哥大学
* 时间:2025年3月14日 16:30-17:30
* 地点:理学院一楼数学系M1001报告厅

## 主讲人简介
Prof. May is a professor at the University of Chicago. He is a world-famous algebraic topologist and a member of the inaugural class of Fellows of the American Mathematical Society. He has made foundational and groundbreaking contributions to many subjects in algebraic topology and homotopy theory, such as computation of the stable homotopy groups of spheres, the May spectral sequences, theory of generalized cohomology operations, infinite loop space theory, equivariant homotopy theory, highly structured ring spectra theory, theory of operads, parametrized homotopy theory, theory of model categories, etc. Prof. May has published more than 100 papers in leading mathematical journals, such as the Annals of Mathematics. He has also written many books in homotopy theory, including a very influential textbook, A concise course in algebraic topology. He had served for many years as the chair of the Department of Mathematics at the University of Chicago, as well as on its graduate admissions and studies committees. Prof. May values education in mathematics very highly. Since 2000, he has been organizing every year the undergraduate summer research program at the University of Chicago, also known as the REU. Until 2024, he has advised more than 60 doctoral students, many of whom have held tenured positions in prestigious universities all over the world. Prof. May has made a significant contribution to the development of algebraic topology in China, especially in educating and influencing a younger generation of Chinese algebraic topologists.

## 讲座简介
Equivariant multiplicative infinite loop space theory is interesting, but hard. I will describe definitions in terms of symmetric groups which promise to give an easy new way to build the theory. As I shall describe, the main obstruction to making the new theory rigorous is the elementary categorical combinatorics of symmetric groups.

## 海报链接
![](https://gtimg.liziwl.cn/post-img/2025-03-14T16-30-00_Prof._Peter_May.jpg)
 No newline at end of file
+19 −0
Original line number Diff line number Diff line
# 《南科大讲堂第395期》邵峰:Gasdermin家族介导的焦亡:从抗菌到抗肿瘤免疫

* 题目: Gasdermin家族介导的焦亡:从抗菌到抗肿瘤免疫
* 主讲人:邵峰 @ 北京生命科学研究所(NIBS)资深研究员、副所长,清华大学讲席教授,中国生物化学与分子生物学学会理事长
* 时间:2025年3月19日 10:00-11:30
* 地点:理学院1142报告厅

## 主讲人简介
邵峰博士现任北京生命科学研究所(NIBS)资深研究员、副所长。1991-1996年就读于北京大学化学系,2003年获得美国密歇根大学博士学位。2004年任哈佛医学院达蒙·鲁尼恩癌症研究基金会博士后研究员。2005年回国进入北京生命科学研究所工作,先后担任研究员、高级研究员、资深研究员。自2020年起受聘为清华大学讲席教授,现任中国生物化学与分子生物学学会理事长。

邵峰院士的研究聚焦于病原菌与宿主免疫系统的相互作用。他系统鉴定了多个细菌模式分子的胞内受体,包括识别细菌脂多糖LPS的caspase-11/4/5及识别LPS生物合成中间产物ADP-heptose的ALPK1受体。其团队发现gasdermin-D蛋白经caspase-1/4/5/11切割后引发细胞焦亡,这一机制在脓毒症等炎症性疾病中起关键作用。该研究确立了gasdermin家族成孔蛋白的核心地位,将细胞焦亡重新定义为由gasdermin介导的程序性坏死。

邵峰院士累计发表SCI论文百余篇,总被引约6万次,H指数达76。其成果获未来科学大奖、William B. Coley基础与肿瘤免疫学杰出研究奖、HHMI国际青年科学家奖、美国蛋白质学会Irving Sigal青年科学家奖等重要奖项。现为中国科学院院士、德国国家科学院(Sciences Leopoldina)院士、欧洲分子生物学组织(EMBO)外籍会士、美国微生物科学院会士及中国医学科学院学部委员。

## 讲座简介
经典和非经典炎症小体通过切割gasdermin D (GSDMD)来诱导焦亡。虽然caspase-1加工IL-1β和IL-18使其成熟,但迄今尚未明确发现caspase-4/5/11针对的细胞因子靶标。本研究表明,活化的人caspase-4能够在体外以及细菌感染过程中(主要发生在上皮细胞中)直接且高效地加工IL-18,而小鼠的caspase-11则不具备这一功能。caspase-4/前体IL-18复合物的晶体结构揭示了一种二元底物识别机制,其中包括一个独特的外部位,该位点能结合由前肽和切割位点后序列共同形成的特定结构。caspase-11在结构上的偏差解释了其无法靶向前体IL-18的原因,而通过合理设计的突变则可以恢复这一功能。我们还发现,在脑血管内皮细胞中,由caspase-4/11激活的GSDMD(而非TLR4诱导的细胞因子)介导了血脑屏障(BBB)在应对循环中的LPS或LPS诱导的败血症时的破坏。电子显微镜观察到受损BBB的超微结构变化,包括焦亡的内皮细胞。将活性GSDMD导入脑血管内皮细胞能够绕过LPS刺激,从而打开BBB。在CASPH4人源化小鼠中,克雷伯氏肺炎杆菌感染导致BBB破坏,而这种破坏可以被在脑血管内皮细胞中表达的GSDMD中和纳米抗体所阻断。这些发现共同改变了我们对于非经典炎症小体介导的抗菌防御和败血症的认识。最后,我将讨论焦亡在激发肿瘤免疫中的作用。具体来说,细胞毒性淋巴细胞通过焦亡杀死GSDMB阳性细胞,这一过程由颗粒酶A对GSDMB的切割介导。IFN-γ上调GSDMB的表达,并促进包括CAR-T/TCR-T细胞介导在内的癌细胞焦亡。

## 海报链接
![](https://gtimg.liziwl.cn/post-img/2025-03-19T10-00-00_%E9%82%B5%E5%B3%B0.jpg)
 No newline at end of file
+15 −0
Original line number Diff line number Diff line
# 《SUSTech Lecture Series 第396期》周孟初 教授:推动知识蒸馏技术发展的多学科方法

* 题目: 推动知识蒸馏技术发展的多学科方法
* 主讲人:周孟初 教授 @ 新泽西理工学院电气与计算机工程系
* 时间:2025年3月19日 16:00-17:00
* 地点:工学院南楼424会议室

## 主讲人简介
周孟初教授于1983年获得南京理工大学控制工程学士学位,1986年获得北京理工大学自动控制硕士学位,1990年获得伦斯勒理工学院计算机与系统工程博士学位。1990年,他加入新泽西理工学院电气与计算机工程系,现为杰出教授。他的研究领域包括智能自动化、机器人、Petri网、物联网、边缘/云计算、人工智能和大数据分析。他发表了1300多篇学术成果,包括17本书籍、900多篇期刊论文(包括700多篇IEEE Transactions论文)、31项专利和32本书籍章节。曾获美国新泽西理工学院杰出研究奖和奖章、美国洪堡基金会颁发的美国资深科学家洪堡研究奖、IEEE系统、人与控制论学会颁发的Franklin V. Taylor纪念奖和Norbert Wiener奖、新泽西州研究与发展委员会颁发的爱迪生专利奖。他是美国华人科学技术协会终身会员,并于1999年担任该协会主席。他是美国国家发明家科学院(NAI)院士,IEEE、国际自动控制联合会(IFAC)、美国科学促进会(AAAS)、中国自动化学会(CAA)的会士。

## 讲座简介
现有的知识蒸馏研究主要集中在以教师网络为中心的方法上,即先根据教师网络自身的标准对其进行训练,然后将所学知识传授给学生网络。然而,由于教师网络和学生网络结构的差异,前者所学的知识可能并不完全符合后者的需求。受人类教育智慧、自动控制理论和模糊逻辑概念的启发,我们提出了一种以学生为中心的蒸馏(SCD)方法,使教师网络能够根据学生网络的实际需求调整其知识传递。我们基于多种人类教育智慧来实现该方法,例如,教师网络在验证集上识别并学习学生网络所需的知识,然后通过训练集将其传递给后者。为解决学生网络在学习过程中面临的知识不足、难样本学习和知识遗忘等问题,我们引入并改进了控制系统领域的比例-积分-微分(PID)算法,使其能够有效识别学生网络当前所需的知识。此外,我们提出了一种基于课程学习的模糊逻辑策略,并将其应用于所提出的 PID 控制算法,使学生网络能够主动关注难样本的学习。实验结果表明,SCD在计算机视觉任务上优于现有的以教师为中心的方法。

## 海报链接
![](https://gtimg.liziwl.cn/post-img/2025-03-19T16-00-00_%E5%91%A8%E5%AD%9F%E5%88%9D_%E6%95%99%E6%8E%88.jpg)
 No newline at end of file
+15 −0
Original line number Diff line number Diff line
# 《大湾区物理讲堂 2025-13期》于海明:反铁磁绝缘体中的自旋波传输

* 题目: 反铁磁绝缘体中的自旋波传输
* 主讲人:于海明 @ 北京航空航天大学
* 时间:2025年3月20日 10:00-11:00
* 地点:Room P1078, College of Science

## 主讲人简介
于海明,北京航空航天大学 集成电路科学与工程学院 教授、博士生导师。2006年本科毕业于北京大学物理学院。2011年博士毕业于瑞士洛桑联邦理工大学(EPFL)。2011-2014年在德国慕尼黑工业大学从事博士后研究。2014年加入北京航空航天大学Albert Fert研究院从事自旋波电子学研究。2015年获海外高层次人才(青年)项目,现任北京航空航天大学教授。主要研究方向:自旋电子学、磁子学、自旋热电子学。

## 讲座简介
Magnons, the quanta of spin waves, are collective excitations of electron spins in magnetic materials, which do not rely on charge transport, and can thus avoid Joule heating in information processing, making them potential candidates for next-generation low-power spintronic devices. Magnonics [1] conventionally study spin waves in ferromagnetic (e.g., CoFeB) and ferrimagnetic (e.g., YIG) materials. Recently, the study of antiferromagnetic (AFM) spin waves has seen a surge of interests [2]. Hematite (α-Fe₂O₃) is one of the materials where the Dzyaloshinskii-Moriya interaction was first discovered, and may potentially exhibits some unconventional magnetism [3]. Apart from the AFM resonance at THz [4], a low-frequency (~10 GHz) mode was reported in hematite with low magnetic damping comparable to that of YIG (10⁵) [5]. In this talk, I will introduce our observation of long-distance propagation of spin waves in hematite with a high group velocity of up to 22.5 km/s [6]. An exchange stiffness of 1.7 Å is deduced from its dispersion relation. Furthermore, we study the spin-wave propagation in hematite below the Morin temperature, i.e., in its easy-axis phase. We unexpectedly discover a nearly zero-frequency magnon mode, which is theoretically interpreted as gapless magnons within the AFM domain walls [7]. While ferromagnetic magnons are always right-handed, AFM magnons can carry both left-handed and right-handed chirality. Finally, we provide direct experimental observation of AFM magnon interference with alternating chirality, which enables the control over the sign of pure spin current in the antiferromagnetic insulator [8].

## 海报链接
![](https://gtimg.liziwl.cn/post-img/2025-03-20T10-00-00_%E4%BA%8E%E6%B5%B7%E6%98%8E.jpg)
 No newline at end of file
Loading