DSI Distinguished Lecture by George Karypis, Associate Professor, Department of Computer Science and Engineering, University of Minnesota

Friday, September 25, 2015, 11:00am-12:00pm
Hariri Institute for Computing, 111 Cummington Mall, Room 180
Please join us for refreshments at noon

Big Data Research: Methods, Systems, and Applications

We are in the era of “Big Data”, which is loosely defined as the application of data driven approaches to solve problems arising in a wide-range of domains in science, engineering, government, and business. Big Data holds the promise of allowing us to tackle problems at a scale, complexity, and fidelity that was previously impossible, enables us to achieve a deep understanding about the world around us, and revolutionize every aspect of our daily life.

In this talk, I present an overview of some recent work in my laboratory that spans various aspects of “Big Data” research including development of new algorithms, runtime systems, and applications of data analysis methods to emerging areas. On the algorithms side, the talk will focus on methods for nearest-neighbor recommender systems, on methods for analyzing dynamic relational networks towards finding patterns of relational co-evolution, on methods for partitioning and clustering networks on multi-core architectures, and on parallel methods for sparse tensor decomposition. On the systems side, the talk will focus on our work in developing runtime systems to allow the automated out-of-core execution of distributed memory message-passing programs, which provides a framework for solving very large problems on moderate size clusters and still achieve high-levels of computational performance. Finally, on the application side, the talk will present our work on employing “Big Data” approaches to analyze higher education data towards addressing issues related to academic pathways, effective pedagogy, retention, and persistence.

Brief Bio:
George Karypis is a Professor at the Department of Computer Science & Engineering at the University of Minnesota in the Twin Cities of Minneapolis and Saint Paul and a member of the Digital Technology Center (DTC) at the University of Minnesota. His research interests are concentrated in the areas of bioinformatics, cheminformatics, data mining, and high-performance computing, and from time-to-time, he looks at various problems in the areas of information retrieval, collaborative filtering, and electronic design automation for VLSI CAD.

Over the years George Karypis has developed algorithms to solve a variety of problems including dynamic load balancing of unstructured parallel computations, graph and circuit partitioning, protein remote homology prediction and fold recognition, protein structure prediction, recommender systems, data clustering, document classification and clustering, frequent pattern discovery in diverse datasets (transactions, sequences, graphs), parallel Cholesky factorization, and parallel preconditioners.

Karypis’ research has resulted in the development of software libraries for serial and parallel graph partitioning (METIS and ParMETIS), hypergraph partitioning (hMETIS), for parallel Cholesky factorization (PSPASES), for collaborative filtering-based recommendation algorithms (SUGGEST), clustering high dimensional datasets (CLUTO), and finding frequent patterns in diverse datasets (PAFI). In addition, he has developed two web-based servers for clustering gene expression data (gCLUTO) and for predicting the secondary structure of proteins (YASSPP).