Mapping Methods for Efficient Similarity-Based Multimedia Re

Google Tech Talk July 30, 2009 ABSTRACT "Mapping Methods for Efficient Similarity-Based Multimedia Retrieval." Presented by Vassilis Athitsos. Similarity-based retrieval is the task of identifying database patterns that are the most similar to a query pattern. Retrieving similar patterns is a necessary component of many practical applications, in fields as diverse as computer vision, bioinformatics, and speech/audio processing. This talk presents three methods that we have recently introduced for speeding up similarity-based retrieval in multimedia databases. What all three methods have in common is that they map the original multimedia retrieval problem into a much easier problem involving retrieval in a Euclidean/vector space. The first method, called BoostMap, is applicable in cases where we want to find the nearest neighbors of the query under a computationally expensive distance measure, such as dynamic time warping, the earth movers distance, or shape context matching. The second method, called embedding-based subsequence matching, is used to find optimal subsequence matches in databases of strings under the edit distance or Smith-Waterman, as well as in databases of time series under the dynamic time warping distance measure. The third method is useful for efficient retrieval of database vectors that maximize the dot product with a query vector, and is applied to speed up application of boosting-based classifiers in domains with a very large number of classes. Experimental results illustrate the advantages of these methods in several application scenarios, including gesture and sign language recognition, optical character recognition, efficient search in DNA databases, and face recognition. Vassilis Athitsos received the BS degree in mathematics from the University of Chicago in 1995, the MS degree in computer science from the University of Chicago in 1997, and the PhD degree in computer science from Boston University in 2006. In 2005-2006 he worked as a researcher at Siemens Corporate Research, developing methods for database-guided medical image analysis. In 2006-2007 he was a postdoctoral research associate at the Computer Science department at Boston University. Since August 2007 he is an assistant professor at the Computer Science and Engineering department at the University of Texas at Arlington. His research interests include computer vision, machine learning, and data mining. His recent work has focused on efficient similarity-based retrieval, gesture and sign language recognition, shape modeling and detection, subsequence matching for strings and time eries, and efficient classification of a large number of classes.