文：西南财经大学统计学院 发布时间：2016-05-16 浏览次数：2076
主 题：Statistical Learning of Neuronal Functional Connectivity
主讲人：Prof. Chunming Zhang
主办单位：统计研究中心 统计学院 科研处
Chunming Zhang received the B.S. in Mathematical Statistics from Nankai University in 1990, the M.S. degree in Computational Mathematics from Chinese Academy of Sciences in 1993, and the Ph.D. in Statistics from University of North Carolina–Chapel Hill in 2000. She was an Assistant professor (2000-2005), Associate Professor (2005-2010), and full professor (2010--) at the Dept. of Statistics, University of Wisconsin-Madison. She served as Associate Editors for Annals of Statistics (2007–2009), Journal of the American Statistical Association (2011–), and Journal of Statistical Planning and Inference (2012–). She was elected to be “Fellow of the Institute of Mathematical Statistics” (2011) and “Fellow of the American Statistical Association” (2016). She was Program Chair–Elect (2014) and Program Chair (2015), Section on Nonparametric Statistics, American Statistical Association. Her research interests include high-dimensional data modeling and inference, non-parametric and semi-parametric inference, large-scale simultaneous inference, with applications to neuroscience and neuroimaging data, bioinformatics, and econometrics and finance.
Identifying the network structure of a neuron ensemble is critical for understanding how information is transferred within such a neural population. However, the spike train data pose significant challenges to conventional statistical methods due to not only the complexity, massive size and large scale, but also high dimensionality. In this paper, we propose a SIE regularization method for estimating the conditional intensities under the GLM framework to better capture the functional connectivity among neurons. We study the consistency of parameter estimation and model selection of the proposed method. An Accelerated Full Gradient Update algorithm is developed to efficiently handle the complex penalty in the SIE-GLM for large sparse data sets applicable to spike train data. Simulation results indicate that our proposed method outperforms existing approaches. An application of the proposed method to a real spike train data set provides some insight into the neuronal network