題目：Co-Clustering of Nonsmooth Graphons

主講：David Choi教授美國卡內基梅隆大學(xué)公眾政策與信息管理系

時(shí)間：7月7號（周四）上午9：00開(kāi)始

地點(diǎn)：bwin必贏(yíng)唯一官網(wǎng)314教室

Abstract: Theoretical results are becoming known for community detection and clustering of networks; 

however, these results assume an idealized generative model that is unlikely to hold in many settings. 

Here we consider exploratory co-clustering of a bipartite graph, where the rows and columns of 

the adjacency matrix are assumed to be samples from an arbitrary population. This is equivalent to 

assuming that the data is generated from a nonparametric model known as a graphon. We show that 

co-clusters found by any method can be extended to the row and column populations, or equivalently 

that the estimated blockmodel approximates a blocked version of the generative graphon, with generalization

 error bounded by n^{-1/2}. Analogous results are also shown for degree-corrected co-blockmodels 

and random dot product bipartite graphs, with error rates depending on the dimensionality of the latent 

variable space.

個(gè)人簡(jiǎn)歷：

David Choi教授于2004年畢業(yè)于斯坦福大學(xué)電子信息工程系獲得博士學(xué)位。后在美國麻省理工學(xué)院Lincoln實(shí)驗室做研究員。2009-2011在哈佛大學(xué)工業(yè)工程系做博士后研究。2011-2012年在加州大學(xué)伯克利分校統計系做兼職教授。2012年進(jìn)入卡內基梅隆大學(xué)公共政策與信息管理系任教。