BYR Achieve · 镜像论坛

题目一：Discovering New Knowledge from the Internet 内容：A large number of natural phenomena, whose laws of governance are yet to be discovered, have in recent years been found to fit nicely with the double Pareto lognormal (DPLN) distribution. These phenomena include, among other things, the size of human settlements, the size of particles, the size of oil fields, the size of world-wide-web files, and the spread and extinguishment of forest wild fires. The DPLN distribution was derived from studying the ups and downs of financial portfolios over a population of investors in the stock market. In this talk we will first try to explain why double Pareto lognormal would make sense for a number of natural phenomena. We will then show that certain aspects of human knowledge, obtained from the Internet, also fit nicely with double Pareto lognormal. We hope that this study will help us obtain insights on the underlying knowledge structure for modeling knowledge generations. Such understanding will in turn help us obtain insights on how human acquire knowledge, which would finally lead us toward an automation of knowledge discovery from the largest storage of knowledge forest that resides in the Internet. This is joint work with Weibo Gong (UMass Amherst), Zheng Fang (UMass Lowell), Benyuan Liu (UMass Lowell), and Xu Yuan (UMass Lowell). 题目二：Constructing Spatial Barriers with Underwater Sensor Networks 内容：Current technologies have made it possible for submarines to thwart standard (active or passive) sonar detections. One viable alternative is to use magnetic or acoustic sensors in close proximity to possible underwater pathways a submarine may pass through. This approach requires deploying large-scale underwater sensor networks to form a spatial barrier. In this talk we will first survey technologies of underwater acoustic networks and 2-dimensional sensor networks. We will then show new results for 3-dimensional sensor networks which are fundamentally different. We first prove that spatial barrier is unlikely to exit in a large 3-dimensional fixed emplacement sensor field where sensor locations follow a Poisson point process. In other words, a path is likely to exit by which an adversary informed of the locations of the sensors can pass through without being detected. We then describe energy conserving approaches to constructing a spatial barrier using mobile nodes so that intruding submarines cannot pass through without being detected. We start by implementing an optimal approach to mapping sensors to grid positions. We then focus on developing an approximate solution for better time efficiency using Auction algorithms and a divide-and-conquer strategy. Our results show that the Auction algorithm produces similar results to the optimal approach at a reduced computational expense, providing an effective approach to constructing an underwater spatial barrier. This is joint work with Benyuan Liu and Stanley Barr. 主讲人：Jie Wang（Department of Computer Science, University of Massachusetts, Lowell）主持人：王柏教授（校学术委员会委员、计算机学院副院长）时间：2010年5月25日（周二）下午3：30-5：30 地点：教三楼136报告厅此讲座为前沿课题讲座，欢迎全校师生踊跃参加。校学术委员会 2010年5月20日

学术讲座通知