DTC Leading Edge Seminar Series
Wednesday, February 20, 2013
3:30 p.m. reception
4:00 p.m. seminar
401/402 Walter Library
View presentation (PDF 4MB)
In this talk I consider the problem of identifying patterns in a data set that exhibit anomalous behavior, often referred to as anomaly detection. Most anomaly detection algorithms involve calculating dissimilarities between data samples using a criterion such as Euclidean distance. However, in many applications involving complex high-dimensional data, there may not exist a single dissimilarity measure that captures all possible anomalous patterns. In such a case, multiple criteria for anomaly detection can be defined. I introduce a novel multi-criteria anomaly detection method using Pareto depth analysis (PDA). PDA uses the concept of Pareto optimality to detect anomalies under multiple criteria. I demonstrate the PDA method for detection of anomalous pedestrian trajectories.
Kevin S. Xu is a Senior Research Scientist in the Computational Intelligence laboratory at 3M. He performs research and development in statistical signal processing and machine learning with applications to security and health care among other fields. He received the BASc in Electrical Engineering from the University of Waterloo (2007) and the MSE and PhD in Electrical Engineering: Systems from the University of Michigan (2012). He was a recipient of the Natural Sciences and Engineering Research Council of Canada Postgraduate Scholarship.