The pervasive use of computers, as well as tremendous advances in the ability to acquire, store and process data, has resulted in a spectacular increase in the amount of data being collected. The availability of high-quality data has in turn lead to many new applications, as well as many new scientific discoveries. Unfortunately, the increased dataset sizes have also revealed scalability problems in many applications. Often these problems are a result of the underlying algorithms not taking the hierarchical structure of modern memory systems into account (and especially the large difference in the access time of main memory and disk). In this talk we will discuss the motivation behind and objectives of the Center for Massive Data Algorithms (MADALGO), established in 2007 by the Danish National Research Foundation. We will especially focus on the centers work on algorithms that minimize the number of accesses to disk - so-called I/O-efficient algorithms - and give examples of massive terrain data applications where the use of I/O-efficient algorithms has led to large practical runtime improvements.

Date:	14:00-14:40, Wednesday, March 25, 2009

Venue:	FIT Building 1-315, Tsinghua University

Title:	MASSIVE Data AlgorithmITCS

Speaker:	Lars Arge

Biography:	Lars Arge is a Professor of Computer Science and Director of Center for Massive Data AlgorithmITCS (MADALGO) funded by the Danish National Research Foundation. He received his Ph.D. in Computer Science from University of Aarhus in 1996 and until August 2004 I was at the Department of Computer Science at beautiful Duke University where He still holds an Adjunct Professor position. He is an elected member of the Royal Danish Academy of Sciences and Letters, and the recipient of an Ole Rømer Scholarship from the Danish National Science Research Council and a Career Award from the US National Science Foundation. His cv has more detail. His main research interests lie in the area of memory-hierarchy efficient algorithms, especially I/O-efficient algorithms for problems involving massive datasets. He has worked on massive dataset problems in many fundamental areas, but much of his work has been on I/O-efficient algorithms and data structures for problems with practical applications in geographical information systems and spatial databases. He has investigated the practical merits of theoretically developed I/O-efficient algorithms in projects such as TPIE, TerraFlow, Stream and TerraStream.

Abstract:	The pervasive use of computers, as well as tremendous advances in the ability to acquire, store and process data, has resulted in a spectacular increase in the amount of data being collected. The availability of high-quality data has in turn lead to many new applications, as well as many new scientific discoveries. Unfortunately, the increased dataset sizes have also revealed scalability problems in many applications. Often these problems are a result of the underlying algorithms not taking the hierarchical structure of modern memory systems into account (and especially the large difference in the access time of main memory and disk). In this talk we will discuss the motivation behind and objectives of the Center for Massive Data Algorithms (MADALGO), established in 2007 by the Danish National Research Foundation. We will especially focus on the centers work on algorithms that minimize the number of accesses to disk - so-called I/O-efficient algorithms - and give examples of massive terrain data applications where the use of I/O-efficient algorithms has led to large practical runtime improvements.