CS Seminar: Counting with Histograms: To Count or Not to Count

CS Seminar: Counting with Histograms: To Count or Not to Count

LISTEN
Title: Counting with Histograms: To Count or Not to Count
Speaker: Egemen Tanin
Date/Time: December 17, 2014 - 12:40-13:30 
Place: FENS G035
Abstract: 
Histograms are a known method for processing data in aggregate form. They are easy to analyze, efficient to store and search, and have many other benefits such as being privacy and security-aware. In particular, counting using histograms is a well-established method. One may prefer this to using explicit identifiers, in particular due to its privacy benefits in today's connected world. In this talk, we will look into problems that occur when one wants to count spatial data using histograms. A popular scenario is when one needs to count number of cars in a region of a city or crossing an intersection. We will see that spatial data, due to the fact that it commonly involves extents--such as a trajectory consisting of a polyline crossing multiple regions in a city--counting is not as simple a task as it initially looks. We will see that distinct counting is a well-established problem for spatial data when histograms are employed, hindering efforts in data mining, privacy, data exchange and many other fronts in computer science. We will then look at some more advanced versions of histograms such as Euler Histograms and how they could be used to address this problem.  
Bio: 
Egemen Tanin is an Associate Professor at the Department of Computing and Information Systems at the University of Melbourne, Australia. Dr Tanin’s areas of expertise include Database Systems and Distributed Algorithms. In particular, he has worked in Spatial Databases for the most part of his academic career. He was the program committee chair for ACM SIGSPATIAL 2011 and 2012 and was also the founding editor of the SIGSPATIAL Special. He has founded and chaired the Australian ACM SIGSPATIAL Chapter. He is currently the Treasurer for ACM SIGSSPATIAL and an associate editor for its journal, ACM TSAS. He continues to serve for other leading organizations related to spatial data management such as SIGMOD and VLDB End., under different roles, e.g., as publicity chair for SIGMOD 2015 which will be held in Melbourne. Dr Tanin has obtained his PhD from the University of Maryland in 2001. He is a graduate of Bilkent University,  1993. He has spent three years during his academic career working for other institutions as such Lawrence Berkeley Labs, NEC, UMIACS, IAI, as well as NASA Goddard, as a visiting researcher. 
Contact: Hüsnü Yenigün