北京师范大学

系统科学会议2020

 

Conference on Systems
Science @BNU 2020

北京师范大学系统科学学院

北京师范大学复杂系统国际科学中心

2021112-13

 

北京师范大学系统科学会议(2020)

Conference on Systems Science @ BNU 2020

为促进教师和学生的学术交流、推动科学研究工作,加速凝练学科方向,落实“一体两翼”发展战略,推进系统科学的学科建设,系统科学学院与复杂系统国际科学中心共同决定举办北京师范大学系统科学会议2020。会议采取大会报告和学生墙报方式进行交流。

 

   :

陈晓松 狄增如

程序委员:

王有贵 王大辉    陈清华

墙报委员:

  陈清华 李   樊京芳

组织委员:

王春凤  李小萌 王梦琦  李维双

 

报告日程:

第一节    112日上午)                                          主持人:王大辉

8:50-9:00

陈晓松

会议致辞

9:00-9:40

樊京芳

统计物理理论在复杂地球系统的应用

9:40-10:20

 

排名聚合的理论与实践

 

第二节    112日上午)                                          主持人:韩战钢

10:40-11:20

 

电突触脑功能初探

11:20-12:00

斯白露

认知智能的计算模型

 

第三节    112日下午)                                           主持人:张柯

14:00-14:40

董雅丽

人类行为复杂性和合作的演化

14:40-15:10

崔鹏碧

基于博弈理论的企业自我审查与政府监管问题研究

15:10-15:40

李小萌

比较优势理论与贸易结构演化分析

15:40-16:10

吴金闪

教育系统科学联通脑科学、学科知识、教和学

 

第四节    113日上午)                                           主持人:吴俊

9:00-9:40

 

科学研究中的团队合作与创新性

9:40-10:20

 

从自动微分到自动建模

 

第五节    113日上午)                                           主持人:樊瑛

10:30-11:00

韩战钢

集群行为与集群智能

11:00-11:30

 

生物复杂系统动力学

11:30-12:00

 

基于果蝇微观神经连接的结构和功能研究

 

第六节     113日下午)                                         主持人:李辉

14:00-14:30

武振伟

非晶态原子连接度与弛豫动力学

14:30-15:00

史贵元

复杂网络上非线性动力学的临界点

15:00-15:30

吴蕊洁

k核分解过程的临界现象

15:30-16:00

 

有向网络的双曲嵌入

16:00-16:10

狄增如

总结

112日,会议地点科技楼B608Zoom在线会议ID670 7298 0398 (Password123456)

113日,会议地点科技楼B608Zoom在线会议ID643 4136 4179 (Password123456)


 

报告摘要

 

樊京芳  教授

题目:统计物理理论在复杂地球系统的应用/ Statistical physics approaches to the complex Earth system

 

摘要:全球变暖、极端气候事件、地震及其伴随的社会经济灾难对我们人类可持续发展构成了严峻的挑战。然而,由于地球系统本身的复杂结构以及存在着众多非线性相互作用,使得人们对于上述灾难性事件的理解,尤其是预测方面变得困难重重。这也是科学界和公共政策的决策者极为关注的话题之一。本次报告我将会对有关统计物理学的基本概念和理论,例如临界现象、网络理论、渗流、临界点分析和熵等,以及如何将其应用于地球复杂系统进行阐述和回顾。

Global warming, extreme climate events, earthquakes and their accompanying natural disasters pose significant risks to humanity. Yet due to the nonlinear feedback, multiple interactions and complex structure of the Earth system, the understanding and in particular the predicting of such disruptive events represent formidable challenges for both scientific and policy communities. During the past years, the emergence and evolution of Earth system science has attracted much attention and produced new concepts and frameworks. Especially, novel statistical physics and complex networks-based techniques have been developed and implemented to substantially advance our knowledge for a better understanding of the Earth system. I will present a brief review on the recent scientific progress in the development and application of how combined statistical physics and complex systems science approaches can beapplied to complex Earth systems.

 

樊京芳博士,北京师范大学系统科学学院副教授,德国波茨坦气候影响研究所(PIK)访问教授。2014年博士毕业于中国科学院理论物理研究所。曾在以色列巴伊兰大学Shlomo HavlinPIKJürgen KurthsHans Joachim Schellnhube教授组分别从事博士后、助理教授工作。美国哈佛大学、波士顿大学访问学者。202010月入职北京师范大学系统科学学院,研究方向为统计物理,气候系统,网络理论,地震系统和金融市场预测等。以第一作者或通讯作者在Nature PhysicsPhysics ReportsPNAS GRL等国际知名期刊发表论文30余篇。

 

 

 

吴俊  教授

题目:排名聚合理论与实践

 

摘要:排名作为一种重要的评价手段广泛存在于大学排名、学生评教、人才选拔、品牌评估等领域。本报告首先介绍了排名以及排名聚合的基本概念与经典方法,进而介绍了本团队提出的基于竞争图的排名聚合方法及在多个领域的实践,最后对排名聚合未来发展进行展望。

 

 

吴俊,北京师范大学自然科学高等研究院复杂系统国际科学中心教授、博士生导师,国防科技大学、伦敦帝国理工大学联合培养博士,加州大学戴维斯分校访问学者,中国系统工程学会青年科技奖获得者、湖南省杰出青年基金获得者,全国百篇优秀博士学位论文(提名)、全军优秀博士学位论文获得者,入选教育部新世纪优秀人才支持计划、国防科技大学首批青年拔尖人才计划,荣立三等功一次。担任国内外十余家权威学术期刊审稿人、多个国内外学术会议主席,中国工业与应用数学学会复杂网络与复杂系统专委会委员、中国管理科学与工程学会理事。长期从事复杂网络与大数据分析研究,特别是在复杂网络抗毁性研究领域取得了突出的学术成绩,相关成果已经在军事、信息、生物等多个领域得到应用。获得军队科技进步二等奖1项、教育部科技进步二等奖1项、湖南省自然科学三等奖1项。主持或参与国家自然科学基金、武器装备预研等科研项目10余项,出版学术专著1部,发表学术论文100余篇,其中第一作者或通信作者SCI论文30余篇。


 

 

 

李岩  教授

题目:电突触脑功能初探

 

摘要:电突触是一种通过缝隙联接来实现信息传递的神经联接方式。与哺乳动物相似,果蝇中有多种联接蛋白形成半通道,两个半通道对接可以直接、快速地传递带电离子、信号分子等小分子。这样的电突触联接网络对于大脑信息处理来说有何种功能呢?我们的研究显示,果蝇蘑菇体中存在电突触联接,对飞行中的果蝇实现视觉学习是必需的。最新研究报道果蝇脑中也存在类似脑电波的振荡,我们在蘑菇体中观察到这种振荡频率与睡眠状态相关,而缺失电突触蛋白则影响这种振荡,也影响睡眠。结合行为学、活体成像与计算模型,我们将逐步揭开电突触在大脑中的功能。

 

 

李岩,1998年毕业于北京大学生命科学学院,2003年获中国科学院生物物理研究所神经科学理学博士,随后在美国加州大学旧金山分校和美国西北大学神经病理学习先后进行博士后和研究助理教授工作。2010年由中科院百人计划引进回国,任生物物理研究所研究员、脑与认知国家重点实验室课题组长、中国科学院大学岗位教授。以果蝇为模式生物长期从事神经生物学研究,在认知、行为调控的神经环路和关键因子,及神经系统发育和疾病等领域开展了深入的、原创性的工作,研究成果发表在Nat Commun, PNAS, J Neurosci, Elife等国际杂志上,多次在GRC, EMBO Conference等国际前沿会议和全国性会议中做大会报告。承担国家自然基金委重大专项、重点项目、国际合作项目、中科院先导和交叉创新团队等重大科技项目。


 

 

 

斯白露  教授

 

题目:认知智能的计算模型

 

摘要:感知、认知与决策的神经机制是脑科学研究的核心问题。脑接收感知信息,整合多个脑区和多种神经元的输入,形成对外部环境的认知是产生智能行为的关键过程。模拟和理解多脑区神经环路在认知任务中的信息处理机制对揭示神经系统的复杂性具有重要意义。在这个报告中,我将介绍认知智能的计算模型的一些进展,并讨论对类脑人工智能的启示。

 

 

斯白露,北京师范大学系统科学学院教授。19992002年分别在北方工业大学和中国科学院计算技术研究所获计算机及应用专业工学学士和硕士学位。2007年获得德国不来梅大学理论神经物理专业博士学位。20082013年期间,先后在意大利国际高等研究院认知神经科学部和以色列魏茨曼研究院神经生物学系从事博士后工作。201312月至20189月任中国科学院沈阳自动化研究所机器人学国家重点实验室研究员。主要研究领域包括类脑智能、集群机器人。代表性成果包括记忆神经环路的计算理论、类脑导航系统等。


 

 

 

董雅丽  博士

图片 1

题目:人类行为复杂性和合作的演化

 

摘要:合作行为在人类和动物群体中广泛存在。随着社会分工体系越来越复杂和经济的发展,个体、组织和国家间的合作也更为重要。2005年,Science杂志在创刊125周年之际,公布了人类亟待解决的125个重大科学问题,其中合作的演化是最为重要的25个问题之一。然而在大多数情况下,个体与群体利益之间会发生冲突,囚徒困境和公共品博弈是研究个体和整体利益最大化之间矛盾的经典博弈模型。奖励和惩罚等激励机制是促进合作的主要方式。传统的激励机制研究大多是基于传统经济学理论,即假设个体是完全理性的,但在现实和实验中发现大多数个体是有限理性的,会表现出从众、风险和损失厌恶、决策误差等。因此如何描述真实个体的行为模式,并以此为依据来设计和分析激励对合作的促进作用很重要。本报告将通过行为实验和演化博弈理论,来研究人类行为的复杂性,并以此为基础分析不同激励机制对合作的影响。

 

 

董雅丽,北京师范大学系统科学学院讲师,研究兴趣为通过结合行为实验,社会交互、经济行为和气候系统大数据,以及理论建模和数值模拟的方法,来研究人类复杂行为系统、气候系统和经济系统,以及三个系统之间的相互影响。长期从事行为经济和气候博弈的研究,特别是在促进合作行为的激励机制研究方面取得了一定的研究成果,相关成果发表在Proceedings of the Royal Society BApplied Mathematics and ComputationThe Singapore Economic Review等国际期刊上。


 

 

 

崔鹏碧  副教授

题目:基于博弈理论的企业自我审查与政府监管问题研究

 

摘要:经济监管研究是经济学中一个非常重要且关键的研究课题。其主要包括两个核心模式:政府从上到下的监管和企业的自律监管。基于实证数据的统计结果,我们注意到两种监管控措的执行度深受行业中违规企业规模或比例的影响。基于实证数据的统计结果,我们在公共博弈框架下通过引入两种惩罚措施即背叛者驱动的惩罚与合作者驱动的惩罚来系统地研究了这个问题。通过对近似理论和基于个体的数值模拟方法,得到了与以往经济学研究相一致的结论:那就是在两种监管共同作用时系统的运行效率更高,更准确地来说在整个过程当中,企业自律监管起着主要作用,信息越对称这种主导作用越强。同时也意外发现,对于两种监管来说,都存在一个中间最优的反馈灵敏度即监管不能太迟钝也不能太灵敏或迅速;所处系统具有小世界结构特征时,系统效率或合作水平达到最高,部分回答了世界社会经济系统往往具有小世界性的原因。该工作的意义在于首次在博弈论框架下研究了政府监管调控和企业自律监管对于经济系统运行所带来的影响,经济监管问题是一个很宏大且关键的研究课题,亟需更深入系统的研究,该工作无疑提供了一个可能的理论研究思路,具有一定借鉴意义。

 

 

崔鹏碧,兰州大学物理学学士,以本硕博就读于兰州大学物理学院,并于2015年获理论物理学博士学位。2015-2019年进入成都电子科技大学计算机科学与工程学院从事博士后研究。2016-2017年获国家留学基金委支持,赴意大利国家研究委员会复杂系统研究所(ISC-CNR)进行合作研究。2019-2020年西北工业大学航天学院副教授。2020年月至今以副教授职称就职于北京师范大学珠海校区自然科学高等研究院复杂系统国际研究中心。研究方向包括:1、复杂系统上的传播扩散动力学行为研究:利用统计物理理论、非线性物理结合传播动力学方法系统研究复杂系统上的耦合传播行为,包括传染病、知识或技术创新的传播。2、 复杂系统中的合作协同行为的研究:利用统计物理、博弈论结合系统理论就社会经济系统或工程集群中的合作协同行为展开研究。


 

 

 

李小萌  博士

题目:比较优势理论与贸易结构演化分析/The universal pathway to commodity structure upgrading in global trade evolution

 

摘要:经济增长是通过不断提升生产和出口商品的技术含量及附加值来实现的。但是,随着全球贸易的发展,蕴含着多边贸易关系和庞杂商品种类的贸易网络具有日趋复杂的拓扑结构和演化特征,仅仅关注部分国家、特定地区,或只讨论少数商品是不全面、也不科学的。本文尝试从全球经济增长的角度对国际贸易关系及演化特征进行量化分析。通过理论模型及实证检验,我们在全球贸易演进中发现了商品比较优势演化的统计特征,以及贸易结构升级的一般性路径。比如,我们验证了低技术水平(或亚线性增长)的商品更易具有比较优势,而随着贸易规模的增长,高技术水平(或超线性增长)的商品在经济达到一定规模后会发生逆转,这种结构性跃迁集中出现在经济规模区分显著的两类群体内部。而对于单个国家来说,出口结构升级则多出现于GDP排名在前50-70%的群体中。

 

 

李小萌,博士毕业于北京邮电大学管理科学与工程专业。专注于计算社会科学相关领域,近期尝试用系统科学的方法量化分析社会经济系统中普遍存在的要素流动现象,从理论上研究要素流动的驱动及障碍、统计特征及演化规律,并将其用于分析人口迁移、国际贸易等领域的实证问题。主持国家自然科学基金一项,发表SCISSCI论文6篇。

 

 

 

 


 

 

 

 

吴金闪  教授

题目:教育系统科学联通脑科学、学科知识、教和学

 

摘要:教和学的研究需要贯通学科研究专家、学科教和学的专家、行为和脑活动的研究者、教育研究者。那,怎么贯通呢?靠学科概念网络这个工具和系统科学的思想。教育系统科学研究中心就是要把这样的思想和工具用于教和学以及和教和学适配的教育管理教师培训等方面的研究。在这里,我会展示一下已经完成的一些工作和正在开展的一些工作,以及计划开展的一些工作。让我们通过问题和数据驱动、实验验证的研究来做真正帮助老师教得更好,帮助学生学得更好的教育研究,携手同行,改变教育,让世界更美丽。

 

 

吴金闪,北京师范大学系统科学学院教授,毕业于The University of British Columbia和北京师范大学。研究工作涉及量子输运(非平衡统计)、量子力学基本问题、量子博弈、博弈论、投入产出分析、科学学、理解型学习,系统科学。一直在实践具有学科基本理论性质的、解决实际问题的、融合学科边界的研究和教学。


 

曾安  副教授

题目:科学研究中的团队合作与创新性

 

摘要:团队合作是当今科学研究中的一个典型特征。已有的研究揭示了团队规模与团队创造力的密切关系。然而,对于团队内部的组成结构,相关研究尚不深入。我们通过分析团队成员的历史发表数据提出了一种基于网络的团队新旧程度量化指标。我们发现,新团队发表的文章在原创性和多学科影响力方面比旧团队更有优势,且这种效应在大规模团队中更加明显。此外,我们发现新团队成员比例比新合作关系比例能更好的对应团队在原创性和多学科影响力方面的表现。最后,我们讨论了团队成员所处的科研生涯阶段与团队创造力之间的关系。本研究不仅在理论上揭示了团队创造力的又一重要关联指标,并可以在实践上指导科学家建立高效的合作关系。

曾安,博士毕业于瑞士弗里堡大学物理系,现任北京师范大学系统科学学院副教授。多年从事复杂网络,科学学和信息过滤的研究,取得了一些有影响力的成果,以第一或通讯作者在国际重要期刊上发表论文80余篇,总引用2000余次,H指数为24。主要论文发表于Nature CommunicationsPhysics Reports等期刊。主持完成两项国家自然科学基金项目。2019年入选仲英青年学者,同年获评北京师范大学励耘青年教师。


 

 

 

张江  教授

题目:从自动微分到自动建模

 

摘要:自动微分技术通过将符号计算与数值计算融合,以实现复杂过程的自动梯度计算,从而构成了深度学习突飞猛进的基础。通过融合重参数化技术,我们实现了基于自动微分的图组合优化问题求解。进一步,我们将该技术用于复杂系统的自动建模,即自动推断系统的相互作用结构和动力学过程,并用控制实验来检验模型是否学到真实的动力学。最后,讲座展望了复杂系统自动建模的未来发展。

 

 

张江,北京师范大学系统科学学院教授,集智俱乐部创始人,集智学园(北京)科技有限公司创始人兼董事长,曾任腾讯研究院特聘顾问。共发表论文SCI论文二十余篇,多篇文章发表在包括Nature CommunicationsNature Machine Intelligence, Scientific ReportsPhysical Review EJournal of Theoretical Biology等国际知名刊物上。主要关注领域:复杂系统分析与建模、复杂网络与机器学习、计算社会科学等。


 

 

 

韩战钢  教授

题目:集群行为与集群智能

 

摘要:自然界,生物展现了众多迷人的集群行为。蚁群的合作,鱼群的旋转群游,鸟群躲避天敌等,都具有大规模集群在没有全局信息共享,无全局规划情况下,出现合作,协同的特性。构建唯像模型与动力机制模型对观测现象和实验中出现的现象进行解释,并研究其相变、对称破缺等性质,探究其简单局部相互作用所引致的整体复杂宏观现象,是系统科学复杂性研究关心的重要方向。生物集群行为研究对机器人群体的仿生控制有很大启发作用。具有巨大应用前景。

 

 

韩战钢, 北京师范大学系统科学学院教授,副院长,校系统分析与集成实验室主任,联合国教科文组织复杂系统数字校园(UNESCO-UniTween-CS-DC)副主席、亚洲区主席。他现任国务院学位办系统科学学科评议组成员,欧亚系统科学研究会理事。

 

曾分别在比利时布鲁塞尔自由大学(ULB)Solvay研究所和美国加州大学洛杉矶分校(UCLA)进行访问研究。他曾经主持和参与多项国家级科研项目,科研成果受到国际同行高度评价。
 
他现在的科学研究集中于通过实验和模型方法研究蚁群、鱼群和机器人群体的集群行为(collective behavior)。以蚁群和鱼群为科学观测对象,通过机器人群体实现仿生控制。研究包括几类群体的集群行为中信息获取、传播和对集群行为的影响,对称破缺(symmetry breaking)的出现与机制分析,系统处于临界态的实验设计、观测与机制分析,系统追逃行为研究,机器人系统自组织行为实现,机器人系统处于临界态的工程实现。这方面研究有很广泛的各行业各领域应用前景。

 


 

 

 

李辉  教授

题目:生物复杂系统动力学

 

摘要:生命是由分子、细胞、组织等不同层次生命物质所构成。其中,细胞作为蛋白质等生物大分子构成的复杂系统,生物大分子在细胞内的运动是代谢、信号传导生命功能的物理基础,揭示它们之间的关系具有极其重要的科学意义。然而,如何精确观测不同层次生命物质的动态过程、刻画其动力学行为、理解其隐藏的功能机理以及多层次间的相互关联,都亟待研究者解决。为此,我们搭建了单分子荧光动态成像平台,提出了特有的测量细胞内扩散、主动运输的动力学研究方法。在分子、细胞尺度开展了一系列复杂动力学研究,发掘其与细胞功能、结构特征以及外界微环境等方面的内在关联,从物理角度理解真实生命过程。

 

 

李辉,现任北京师范大学系统科学学院教授,博士生导师。2006年和2012年分获山东大学学士学位和中科院物理研究所博士学位。2012年毕业留所工作,历任助理研究员、副研究员,并先后赴牛津大学、哈佛大学、麻省理工学院进行学术访问。2019年调入北京师范大学。

 


 

 

 

张柯  教授

See the source image

题目:果蝇微观认知神经网络的解析与启发

 

摘要:近年来,以超薄连续切片文库和电子显微镜成像技术为基础的微观神经连接组得到了长足的发展。该技术可以展示神经系统在突触水平的连接网络,堪比神经系统的电路图,提供了从底层网络结构出发,解析神经系统运行机制的机会。在此基础上比较和分析生物神经网络和人工神经网络的结构、发育和重塑的异同也成为可能。我们从果蝇视觉和嗅觉两个系统入手,利用微观神经连接组技术建立网络图谱,分别关注快速视觉的神经原理和嗅觉信息的传递与整合问题,希望为自主视觉和认知神经网络构建提供新的思路。同时,由于连接图谱数据规模巨大,传统的神经科学分析方法在面对连接组数据时具有局限性,急需引入系统科学思路和方法。

 

张柯,2002年毕业于南京大学生物系生理学专业获得学士学位,2008年于中国科学院上海生命科学研究院神经科学研究所获得神经生物学博士学位,2008年至2020年于中国科学院上海生命科学研究院神经科学研究所从事研究工作,2020年起就职于北京师范大学珠海校区自然高等研究院复杂系统国际科学中心。曾获得中国科学院卢嘉锡青年人才奖、全国百篇优秀博士论文奖、中国科学院院长特别奖、上海青年科技启明星等奖项。关注动物认知与群体行为机制,研究兴趣包括强化学习与抉择行为的神经机制、认知功能的微观神经联结组、动物社会学习神经基础、视觉驱动的集群原理与仿生智能构建等,努力推动生物智能向类脑智能的转化与迁移。


 

 

 

 

武振伟  副教授

题目:非晶态原子连接度与弛豫动力学

 

摘要:非晶态物质的本质及形成过程是凝聚态物理领域最困难也是最有趣的问题之一。非晶形成过程在原子 结构上不会衍生出人们在传统晶体结构里所熟悉的长程有序性,因此对于此类在自然界中广泛存在的物质形态,至今还没有有效的实验表征手段和理论研究方法。非晶态物质的原子结构及其构效关系的研究是凝聚态物理和材料科学等众多研究领域所关注的热点问题之一。随着对非晶态物质物性研究的深入,人们逐渐意识到非晶态物质中原子中程序对系统性质的重要影响,建立以中程序为基础的结构-动力学关系对于理解玻璃及玻璃转变的本质起着重要的作用。本次报告将简要介绍基于图论提出的原子局域连接度这一新的结构序参量在液体和玻璃的结构及构效关系研究中的应用。新的结构序参量从过去侧重于关注局域原子团簇的种类和分布,转移到更加关注某一类具有特殊对称性的原子的空间连接情况,即更多地尝试从超越原子短程序的角度来建立非晶态物质中的结构动力学关系。新的研究结果表明,局域连接度可与非晶态物质中原子的短时或长时动力学行为、输运方式、以及振动模态等一系列物理性质建立联系。

 

 

武振伟,北京大学力学与工程科学系理学博士,北大优博,北京大学物理学院量子材料科学中心博雅博士后,现任北京师范大学系统科学学院副教授。主要从事非晶态物理与材料领域的研究工作,以计算机模拟结合统计物理、图论、机器学习等研究方法,探索液体及玻璃态物质等典型无序非平衡体系的结构与结构动力学关联,探索非晶态物质构型空间及动力学空间中的拓扑结构及其构效关系。具体研究内容如:非晶态物质原子结构的描述,过冷液体中的动力学异质性、晶化过程、输运方式等行为,玻璃化转变过程中的结构序参量,以及玻璃态物质中塑性事件及其对应本征缺陷研究。

 

 

 

 

 

 

 

 

史贵元  副教授

题目:复杂网络上非线性动力学的临界点

 

摘要:从基因调控开关、个体的生死,到流行病的传播、社会经济的兴衰、城市人口数的增减、生态系统的破坏与修复。我们关心的许多现象都不是简单连续的线性关系,而是有多个稳态,并存在从一个稳态突然跃变到另一个稳态的临界点。更有趣的是,正如冬夏之间是春天,而夏冬之间是秋天一样,当外部条件循环往复变化,系统状态并不是原路返回。简单的非线性动力学模型可以帮助我们理解这些现象背后的机制,并发现这些看似完全不同系统之间统一的规律。 单一个体或群体的非线性动力学问题,可以由一个简单的微分方程描述,作为一个传统的数学问题得到了深入的研究。但由于大数量微分方程组求解的复杂性,对于异质多群体之间存在非线性相互作用问题的研究正方兴未艾。本报告将从非线性动力学视角,探讨对于身边一些现象的理解。并介绍结合复杂网络建模,分析异质性群体之间非线性相互作用问题的最新理论进展。

 

 

史贵元,中国科学技术大学物理学学士,瑞士弗里堡大学物理学博士,博士毕业后在瑞士弗里堡大学和丹麦哥本哈根大学尼尔斯玻尔研究所从事复杂系统研究。现为北京师范大学珠海校区复杂系统国际科学中心副教授。研究兴趣为复杂网络上的相变、复杂系统的非线性动力学等理论问题。

 

 

 

 

 

吴蕊洁  博士

题目:k核分解过程的临界现象

 

摘要:K核分解常用来寻找复杂网络中的重要节点,例如疾病传播中的关键传播者、社交网络的中心人物、经济危机时造成重大影响的国家或公司、维持生态系统稳定最需要保护的物种等等。一个网络的K核是该网络满足如下条件的最大子图:该子图中每个节点都能在这个子图里找到不少于K个邻居。对于任意给定的整数K,要找到一个给定网络的K核,最常用的方法是重复迭代地删掉剩余子网络中邻居数小于K的节点以及和这些节点相连的边。迭代过程的终止有两种情况:1. 所有节点都被删除,即该网络不存在K核;2. 剩余子图满足K核的条件,即没有节点可以再被删除。本报告将介绍,对于大规模网络,K核分解可以看作是一个有一系列离散中间态的相变过程;以及如何求解整个相变过程中的临界现象。

 

 

吴蕊洁,中国科学技术大学物理学学士,瑞士弗里堡大学物理学博士,丹麦哥本哈根大学尼尔斯玻尔研究所博士后。现为北京师范大学珠海校区复杂系统国际科学中心讲师,从事复杂网络理论研究。主要研究成果包括复杂网络上非线性动力学的临界点、k核分解相变过程的理论解、不相等人数二部图匹配的解析解、经济复杂性中Fitness-Complexity算法的收敛性分析等。

 

 

 

 

 

 

 

 

 

樊瑛  教授

fan002

题目:有向网络的双曲嵌入

 

摘要:本讲座主要介绍适用于有向网络的双曲空间嵌入模型及其应用。该模型根据有向网络的二分结构和多路复用节点信息,展示了有向边在双曲空间中的连接模式。有向网络双曲嵌入的关键在节点双曲坐标的估计,为此给出了非对称流行度-相似度优化(APSO)方法,通过极大似然估计推断出节点在双曲空间的位置。

 

 

樊瑛 现为北京师范大学系统科学学院教授、博士生导师。中国系统工程学会副秘书长、常务理事。研究方向为复杂性理论及其在各领域中的应用,目前主要关注复杂网络相关研究,并取得了一系列科研成果。

 

 

 

 

 

 


 

墙报展示

 

No.2020.001: Jiaxu Hou

Model-Free H∞ Optimal Tracking Control of Constrained Nonlinear Systems via an Iterative Adaptive Learning Algorithm

 

In this paper, an H∞ optimal tracking controller for completely unknown discrete-time nonlinear systems with control constraints is obtained by using an iterative adaptive learning algorithm. An augmented system is established by integrating the tracking error system and the reference trajectory. As an identifier of the unknown systems, a neural network (NN) is introduced with asymptotic stability of the estimation error. An action–disturbance–critic NN structure is proposed to implement the iterative dual heuristic programming algorithm with convergence guarantee of the costate function and the control policy. Simulation results and comparisons are provided to illustrate the superior performance of the designed optimal tracking controller.

点击查看

 

No.2020.002: Siyu Huang

Effects of Regional Trade Agreement to Local and Global Trade Purity Relationships

 

In contrast to the rapid integration of the world economy, many regional trade agreements (RTAs) have also emerged since the early 1990s. +is contradiction has encouraged scholars and policymakers to explore the true effects of RTAs, including both regional and global trade relationships. +is paper defines synthesized trade resistance and decomposes it into natural and artificial factors. Here, we separate the influence of geographical distance, economic volume, and overall increase in transportation and labor costs and use the expectation maximization algorithm to optimize the parameters and quantify the trade purity indicator, which describes the true global trade environment and relationships among countries. +is indicates that although global and most regional trade relations gradually deteriorated during the period 2007–2017, RTAs generate trade relations among members, especially contributing to the relative prosperity of European Union (EU) and North American Free Trade Agreement (NAFTA) countries. In addition, we apply the network to reflect the purity of the trade relations among countries. +e effects of RTAs can be analyzed by comparing typical trade unions and trade communities, which are presented using an empirical network structure. +is analysis shows that the community structure is quite consistent with some trade unions, and the representative RTAs constitute the core structure of international trade network. However, the role of trade unions has weakened, and multilateral trade liberalization has accelerated in the past decade. +is means that more countries have recently tended to expand their trading partners outside of these unions rather than limit their trading activities to RTAs.

点击查看

 

No.2020.003: Yuanxiang Jiang

Characterizing dissimilarity of weighted networks

 

Measuring the dissimilarities between networks is a basic problem and wildly used in many fields. Based on method of the D-measure which is suggested for unweighted networks, we propose a quantitative dissimilarity metric of weighted network (WD-metric). Crucially, we construct a distance probability matrix of weighted network, which can capture the comprehensive information of weighted network. Moreover, we define the complementary graph and alpha centrality of weighted network. Correspondingly, several synthetic and real-world networks are used to verify the effectiveness of the WD-metric. Experimental results show that WD-metric can effectively capture the influence of weight on the network structure and quantitatively measure the dissimilarity of weighted networks. It can also be used as a criterion for backbone extraction algorithms of complex network.

点击查看

 

No.2020.004: Lingbo Li

Opinion Dynamics on Signed Networks Based on Ising and Potts Model

 

The evolution dynamics of public opinion is a hot issue in complex network research, and signed networks can describe amicable and antagonistic relationships in complex real-world systems accurately. Exploring dynamic processes such as opinion interactions on singed networks can help us understand the evolution process in the real world, but few people have paid attention to it. Previous methods for opinion diffusion cannot be applied to signed network directly, which ignore the important information contained in the negative edges. In this work, we use statistical physics approaches to understand the influence of the network topology on emergent properties such as the evolution of opinion, and apply such model to real socioeconomic system to solve practical problems. • In part I, we apply the Ising Model to signed networks with different structures. The results reveal the importance role of the negative edges. • In part II, we concentrate on the Potts Model on signed networks and use it to study the relationship between economic development and political election. We provide a simple election prediction model that requires only readily available economic data.

点击查看

 

No.2020.005: Meng Li, Yuanxiang Jiang, Zengru Di

Charactering the reputation of evaluators by the vector in feature space of objects

 

Reputation is especially important in online systems and has attracted much attention. How to evaluate user reputation comprehensively based on the rating information of the user-object bipartite networks is a fundamental problem due to the existence of malicious ratings and spamming attacks. Most previous ranking-based methods have been proposed for addressing this problem. In this paper, we propose a reputation evaluation algorithm in terms of the classification of objects. More specifically, the objects are classified into several categories according to the structure information obtained from the one-mode projection onto object of bipartite network. The reputation vector of a user represents his/her reputation for the objects in different categories. In addition, the user is assigned an appropriate attribute according to his/her reputation vector through K-Means. Results on both artificial rating data and real rating data suggest that the presented method has better performance than the traditional correlation-based ranking method in both accuracy and robustness.

点击查看

 

No.2020.006: 林国政

生物集群中状态转换过程的研究及应用

 

生物集群行为广泛存在于自然界中,例如细胞的集群迁移,蚁群搬运食物,鸟类集群躲避天敌等等,这些由大量个体通过微观局部的相互作用组成的系统能够涌现出复杂的宏观行为。这吸引了生物、物理、集群智能、复杂性等学科科学家的关注。本研究以理论和实验相结合的方法,探索集群运动中状态转换过程的机制。在理论上,建立能够整合“velocity-based”“position-based”两类机制的新模型在实验上;在实验上,设计和开展可控的鱼群实验,揭示群体从有序到无序再到有序这一过程的特征和机制。最后,将理论和实验的成果应用于机器人和无人机的集群控制中。

点击查看

 

 

No.2020.007: Jing Liu

Tensor networks for unsupervised machine learning

 

"The central task in unsupervised machine learning is to model joint distribution of the data. In recent years, tensor networks have been extended to this task, however only as a proof of principle, because the performance is much worse than the models based on neural networks. In this work, we present the Autoregressive Matrix Product States (AMPS), a tensor-network-based model combining matrix product states and autoregressive modeling of joint probability distribution. Our model enjoys exact calculation of normalized probability and unbiased sampling, as well as a clear theoretical understanding of expressive power.

We apply our model to two machine learning tasks, the generative modeling of standard handwritten digits, and reinforcement learning of variational free energy in graphical models. The experimental results show that the proposed model significantly outperforms existing statistical physics inspired models such as the Restricted Boltzmann machines, as well as the tensor network based models using matrix product states, tree tensor networks, and locally purified states, and is competitive with the state-of-the-art neural network models. Our model paves the way to develop machine learning models competitive to deep neural networks based on tensor networks and quantum circuits."

点击查看

 

No.2020.008: Zhaofan Liu,Yin Yan , DaHui Wang

Experience-dependent representation of visual contour stimulus in primary cortex

 

Perceptual training is known as performance improvement in perceptual tasks, which is known associated with interaction of multiple cortex areas. However, little is known about that the changes of representation of contour patterns during training. By using implanted multi-electrode arrays, we find during perceptual learning, the responses of different behavior (different saccade directions) are diverging while responses of same behavior (same saccade directions) are converging together. This indicating discriminating the contour pattern may due to the category learning: The representations of contour patterns are compressed while representations of noise pattern and contour pattern are exaggerated. Secondly, we find contour input stabilized the neural responses and training enhance the stability. Moreover, the noise correlation is decreasing reflecting the shared spontaneous is suppressing meanwhile the dimensionality increasing reflecting suppression of redundant information.

点击查看

 

No.2020.009: Shuihan Qiu , Kaijia Sun and Zengru Di

Collective dynamics of neural networks with sleep-related biological drives in Drosophila

 

The collective dynamics of the brain as a result of sleep-related biological drives in Drosophila are investigated in this paper. Based on the Huber-Braun thermoreceptor model, the conductance-based neurons model is extended to a coupled neural network to analyze the local field potential (LFP). The LFP is calculated by using two different metrics: the mean value and the distance-dependent LFP.  The distribution of neurons around the electrodes is assumed to have a circular or grid distribution on a two-dimensional plane. Regardless of which method is used, qualitatively similar results are obtained that are roughly consistent with the experimental data. During wake, the LFP has an irregular or a regular spike. However, the LFP becomes regular bursting during sleep. To further analyze the results, wavelet analysis and raster plots are used to examine how the LFP frequencies changed. The synchronization of neurons under different network structures is also studied. The results demonstrate that there are obvious oscillations at approximately 8 Hz during sleep that are absent during wake. Different time series of the LFP can be obtained under different network structures. As the number of coupled neurons increases, the neural network becomes easier to synchronize, but the sleep and wake time described by the LFP spectrogram do not change. Moreover, the parameters that affect the durations of sleep and wake are analyzed.

点击查看

 

No.2020.010: 宋诗佳,李汉东

删失GP-DCS-VaR模型以及应用研究

 

本文提出了一种改进的参数化VaR预测模型,即删失广义帕累托分布下的动态得分驱动VaR(删失GP-DCS-VaR)模型。模型首先基于日内高频数据,利用删失广义帕累托分布(GP)对日内收益的超值进行拟合,然后使用DCS模型对GP分布的动态条件参数进行建模和估计,在此基础上,本文分别介绍了利用多个日内收益分布拟合日收益分布的参数法和自举法,得到日收益的GP分布,从而得到日VaR的预测值。最后,我们利用中国股票市场的数据进行实证分析,样本外预测结果表明GP-DCS-VaR对极端收益的覆盖能力强于已实现的GARCH-VaR,模型可以很好预测中国股票市场的VaR测度,展现了其有效性。

关键词:VaR; 删失GP分布; POT; DCS;回测

点击查看

 

No.2020.011: Ningning Wang

Effect of Migration on Epidemic Spreading in Heterogeneous Metapopulation

 

Migration plays a crucial role in epidemic spreading, and its dynamic can be studied by metapopulation model. Generally, the uniform mixing is a common assumption in the metapopulation, and this assumption implies that contact network is homogeneous and contact exists between any pair of individuals. While, the individual migration and daily encounter can make the contact network keep in dynamic evolution state. Hence, the dynamic network is more suitable to study the effect of migration. In this paper, a population, its individuals coupled by heterogeneous network, is called heterogeneous metapopulation. We use the time averaging to transform the dynamic network into a static weighted network, then obtain the topology for epidemic spreading. With the gravity law of migration, we establish the N-intertwined seat SIR model and obtain its basic reproduction number R0. From this quantitative indicator, we can see that the uniform mixing assumption will significantly overestimate the epidemic reproduction. Meanwhile, we also find that the migration isn't the immediate cause, leading to the improvement of epidemic reproduction, but its effect on the network structure. In addition, we find a bistable state in the heterogeneous metapopulation through dynamic node search algorithm, which reveal that the migration would increase the uncertainty risk of epidemic spreading.

点击查看

 

No.2020.012: Zongning Wu, Zengru Di, Ying Fan

An asymmetric popularity-similarity optimization method for embedding directed networks into hyperbolic space

 

Network embedding is a frontier topic in current network science. The scale-free property of complex networks can emerge as a consequence of the exponential expansion of hyperbolic space. Some embedding models have recently been developed to explore hyperbolic geometric properties of complex networks—in particular, symmetric networks. Here, we propose a model for embedding directed networks into hyperbolic space. In accordance with the bipartite structure of directed networks and multiplex node information, the method replays the generation law of asymmetric networks in hyperbolic space, estimating the hyperbolic coordinates of each node in a directed network by the asymmetric popularity-similarity optimization method in the model. Additionally, the experiments in several real networks show that our embedding algorithm has stability and that the model enlarges the application scope of existing methods.

点击查看

 

No.2020.013: Hao Yu, Xuefeng Cui

Reduced impacts of heat extremes from limiting global warming to under 1.5 ℃ or 2 ℃ over Mediterranean regions

 

Heat extremes have a serious impact on humans and on agriculture around the world. As one of the prominent climate changes “hot spots,” the Mediterranean area, especially the eastern portion, is expected to be more vulnerable to heat exposure than other regions due to its high population density and urbanization rate. The Paris Agreement includes the goal of “holding the increase in the global average temperature to well below 2 °C above preindustrial levels and pursuing efforts to limit the temperature increase to 1.5 °C above preindustrial levels”. It is interesting to study how heat extremes would change in the Mediterranean area in a +1.5 °C and +2 °C global warming world and how they would impact humans and agriculture. Based on high resolution climate scenario data from Coordinated Regional Downscaling Experiment Mediterranean (MED-CORDEX), we calculate several heat extreme indices to answer the above questions.

点击查看

 

No.2020.014: Aobo Zhang

Detangling the Multilayer Structure from an Aggregated Network

 

Multiplex interactions are common and essential in real-world systems. In many cases, we can only obtain aggregated networks without detailed information regarding the type of links contained within. Such single-layer networks simplify the structural information and lead to misunderstandings of some properties of real systems. In this case, network splitting, which aims to correctly sepa-rate an aggregated network into two multilayer networks, is a meaningful problem to study. To solve this problem, we propose a simulated annealing-like algorithm based on the link clustering coecient. We verify the validity of this algorithm with several synthetic networks (regular networks and small-world net-works with dierent properties). Overlapping links between layers are also taken into consideration, and we can find that the proposed method is valid even if there is a certain proportion of overlapping links. When we apply the algorithm to inter-national trading networks, it results in accurate splits of dierent layers.

点击查看

 

No.2020.015: Yanmeng Xing, Fenghua Wang, An Zeng, Yin Fan

Solving the cold-start problem in scientific credit allocation

 

An almost universal trend in science today is prominence of ever-increasing collaborative teams. Hence, identifying the relative credit of each collaborator to the co-author work is of extreme significance. Despite that numerous methods have been proposed to address this issue, how to allocate credit to authors in newly published papers is still challenging. To address this cold-start issue, we introduce a credit allocation algorithm based on the co-citing network that captures the coauthors’ shared credit in a multi-author publication. Using the American Physical Society publication data, we validate the method in the scenario of determining the laureates of Nobel-winning papers that are awarded for the achievement.

Here, we perform a series of experiments to demonstrate that the proposed method can be implemented to academic papers in any period after the publication, specially with a significantly higher accuracy and robustness than existing algorithms for new papers. This method allows us to explore universal credit evolution pattern of scientific elites. Importantly, by testing the relation between an author's credit and authorship byline, we find the last authors in papers nowadays are assigned more credit than those in the early days in physics. With collaboration and big team set to dominate the agenda of current science system, our study provides a more effective method for allocating early credit to coauthors in a paper, which can be beneficial to various academic activities including faculty hiring, funding, and promotion decisions.

点击查看

 

No.2020.016: Yan Zhang

Automated Discovery of Interactions and Dynamics for Large Networked Dynamical Systems

 

Understanding the mechanisms of complex systems is very important. Networked dynamical system, that under- standing a system as a group of nodes interacting on a given network according to certain dynamic rules, is a powerful tool for modelling complex systems. However, finding such models according to the time series of behaviors is hard. Conventional methods can work well only on small networks and some types of dynamics. Based on automatic differentiation technique, this paper proposes a unified framework for Automated Interaction network and Dynamics Discovery (AIDD) on various network structures and different types of dynamics. The experiments show that AIDD can be applied on large systems with thousands of nodes. AIDD can not only infer the unknown network structure and states for hidden nodes but also can reconstruct the real gene regulatory network based on the noisy, incomplete, and being disturbed data which is closed to real situations. We further propose a new method to test data-driven models by experiments of control. We at first optimize a controller on the learned model, and then apply it on both the learned and the ground truth models. The results show that both of them behave similarly under the same control law, which means AIDD models have learned the real network dynamics correctly.

点击查看

 

No.2020.017: Hongbing Xia

Approximate Optimal Sliding Mode Tracking Control for Modular Reconfigurable Robots Based on Critic-only Structure

 

In this paper, an approximate optimal sliding mode tracking control (SMTC) strategy is investigated for modular reconfigurable robots (MRRs) through critic-only structure based-adaptive dynamic programming (ADP) scheme. The SMTC is achieved by three parts, i.e., the optimal control of the nominal system, the sliding mode-based iterative control and an adaptive robust term. The sliding mode-based iterative controller suppresses the error caused by the trajectory tracking, and the adaptive robust term is employed to ensure the reachable condition of sliding mode surface. By solving the Hamilton-Jacobi-Bellman equation with the critic neural network only, the sliding mode-based iterative control can be derived. The SMTC strategy can drive the MRR to present achieve a faster control action based on the approximate optimal control. The closed-loop MRR system is guaranteed to be asymptotically stable under the developed SMTC policy. At last, the effectiveness of the presented strategy was validated via the comparative simulation.

点击查看

 

No.2020.018: Fei Song

A social behavior model based on PFC and the computational mechanism analysis of related brain regions

 

In a conventional three-box social task, mice can distinguish different social objects and exhibit a "novelty-approach" phenomenon in behavior. Experiments have shown that this phenomenon is regulated by the Prefrontal cortex (PFC) and its surrounding  closely connected regions. In order to reveal the functions of different brain areas (especially PFC) in this social behavior, we constructed a two-layer RNN network and applied FORCE/full-FORE learning to train the model. According to the current results, the network has successfully demonstrated neuron firing trajectories similar to the real electrophysiological recording results. In addition, by adjusting the feedforward- and feedback-related parameters of the two-layer network, as well as the relevant experimental conclusions, we propose a conjecture: (1) Social object individuals are encoded by the PFC brain region by extracting characteristics and forming related memories. The expected value represented in the PFC primarily drive learning processes; (2) Social object individuals are separated by PFC downstream brain areas (such as thalamus) by assigning intrinsic "value" to different objects. The expected value represented in these region is used for updating choice mechanisms in other regions of the brain. At the same time, the abstract representation information is transmitted to PFC which affects the learning of Characterization.

点击查看

 

No.2020.019: Kai Zhao

Neural circuit mechanism of left-right decision-making behavior in mice in auditory training behavior resolution experiment

 

In the experiment of different acoustic stimuli, the auditory cortex of mice received different frequencies of external acoustic stimuli. Mice trained in auditory resolution behavior were able to distinguish between frequencies and make left or right choices accordingly. The two-photon imaging signals in the mouse brain showed that the sound stimulus information was transmitted from the cortex to the PV neurons in the striatum of the mouse, and the PV neurons had a probabilistic projection to the spiny neuron in the striatum. In the final decision, the substantia nigra neurons in the basal ganglia are directly inhibited by the spiny neuron D1, while the spiny neuron neuron D2 activates SNR through an indirect circuit. We made a neural network model to simulate the behavior of mice, which can give the predicted results. These results reveal an important phenomenon that the striatum and basal ganglia play an important role in cognitive decision-making, and provide ideas for cognitive decision-making and the treatment of brain diseases.

点击查看