EE Systems Seminar

MapReduce is a commonly used framework for executing data-intensive tasks on distributed server clusters. We present ''Coded MapReduce", a new framework that enables and exploits a particular form of coding to significantly reduce the inter-server communication load of MapReduce. In particular, Coded MapReduce exploits the repetitive mapping of data blocks at different servers to create coded multicasting opportunities in the shuffling phase, cutting down the total communication load by a multiplicative factor that grows linearly with the number of servers in the cluster. We will further discuss the tradeoff between the ''computation load'' and the ''communication load" in distributed computing.

Bio: Salman Avestimehr is an Associate Professor at the Electrical Engineering Department of University of Southern California. He received his Ph.D. in 2008 and M.S. degree in 2005 in Electrical Engineering and Computer Science, both from the University of California, Berkeley. Prior to that, he obtained his B.S. in Electrical Engineering from Sharif University of Technology in 2003. He was an Assistant Professor at the ECE school of Cornell University from 2009 to 2013. He was also a postdoctoral scholar at the Center for the Mathematics of Information (CMI) at Caltech in 2008. 

Dr. Avestimehr has received a number of awards, including the Communications Society and Information Theory Society Joint Paper Award, the Presidential Early Career Award for Scientists and Engineers (PECASE), the Okawa Foundation Research Grant, the Young Investigator Program (YIP) award from the U. S. Air Force Office of Scientific Research, the National Science Foundation CAREER award, and the David J. Sakrison Memorial Prize. He is currently an Associate Editor for the IEEE Transactions on Information Theory.