你会为自己未来的研究生生活感到压力吗？

曾颖琪

能分享你对未来的畅享吗[不看][不听]

来自Android客户端

曾颖琪

各位帮忙顶一顶啊哈

金针菇99

我帮帮你，顶上去

新的生活努力吧

博士再说压力，研究生论文又没什么要求，核心而已

曾颖琪

新的生活努力吧 发表于 2019-3-5 15:31
博士再说压力，研究生论文又没什么要求，核心而已

没有啊，我们要SCI啊，我枯了

曾颖琪

金针菇99 发表于 2019-3-5 15:23
我帮帮你，顶上去

谢谢啦！

曾颖琪

日常顶一顶

曾颖琪

顶一顶哈哈

曾颖琪

各位老师同学，

以下是讲座信息，欢迎大家转发都参加。谢谢。

邹勇



学术报告信息表
讲座题目：Causal network inference and data-driven reconstruction of nonlinear dynamics

主讲人：孙杰

开始时间：4月25日（周四）10:00:11:30

讲座地址：物理楼226

报告人简介：Dr. Jie Sun is a tenured Associate Professor of Mathematics at Clarkson University, USA. He is an Editor of Chaos and Associate Editor of Mathematics in Science and Industry. Before joining Clarkson University, Dr. Sun was a postdoc at Northwestern University and Princeton University from 2010 to 2012. Sun obtained his B.Sc in Physics at Shanghai Jiao Tong University in 2006 and PhD in Mathematics at Clarkson University in 2009. His current research focuses on data-enabled science and applications such as causal network inference, information flow, sensor localization and data-driven modeling of nonlinear dynamics and complex systems. Dr. Sun has over 40 publications including several in Physical Review Letters, Physical Review X, and SIAM. Some of his work were featured in Bloomberg and The Guardian.

报告内容简介:

Causal network inference and data-driven reconstruction of nonlinear dynamics

Abstract: Extracting useful information regarding the dynamics and functioning of complex systems is an important task in modern science. A central problem is to infer causal networks from observational data. This problem is challenging especially when the system is high-dimensional and the dynamics is nonlinear and noisy. To tackle this problem, we utilize concepts from information theory and prove that under mild Markov conditions, the causal inputs of a given node form the minimal set of nodes that maximizes causation entropy, a criterion that we refer to as the optimal causation entropy principle. This principle establishes the equivalence between causal network inference and optimization of causation entropy, and guides us to develop optimal causation entropy (oCSE) algorithms for causal network inference.

    In addition to benchmarking the proposed oCSE method on Gaussian stochastic processes on random networks, we present several examples of application including identification of coupling structure in biological circuites, inference of information flow and intercommunication networks in collective animal behavior, structural health monitoring from spatially distributed sensor time series data, and reconstruction of Boolean functions and networks. Finally, we discuss our recent work of “Entropic Regression”, which enables robust data-driven reconstruction of nonlinear dynamical systems under the presence of large observational noise and outliers.