Reception: 3:00 – 3:30pm - Duthie Center 2nd Floor
Seminar 3:30pm
Place: Duthie Ctr, Room 117
Date: Friday, April 16, 2010

Random Sets and Multiple Instance Learning for Pattern Recognition and Machine Learning

Speaker: Dr. Paul Gader
University of Florida

ABSTRACT:
Pattern recognition and machine learning provide a bevy of probabilistic and statistical techniques for solving classification problems. Sometimes, however, these techniques do not work well for processing multi-dimensional signals because the basic assumptions that patterns can be completely identified does not hold. Random sets offer a generalization of standard probabilistic techniques by replacing real number or vector outcomes of random variables with set value outcomes. The theory of random sets provides a natural framework for Multiple instance learning (MIL). MIL is a technique used for identifying a target pattern within sets of data. In MIL, a learner is presented with sets of samples; whereas in standard techniques, a learner is presented with individual samples. In this talk, the notion of random sets and MIL will be discussed, classification algorithms based on these notions will be discussed, and experimental results will be given with several data sets, including landmine detection using ground penetrating radar and hyperspectral image data sets.

BIO:
Paul Gader received his Ph.D. in Mathematics for image processing related research in 1986 from the University of Florida. He has worked as a researcher at Honeywell’s Systems and Research Center and the Environmental Research Institute of Michigan. He has been in Academia since 1991 and is currently a Professor of Computer and Information Science and Engineering at the University of Florida. His research has spanned a number of areas related to image analysis and pattern recognition including applied mathematics, parallel image processing, mathematical morphology, fuzzy and random sets, handwriting recognition, bio-medical image analysis, automatic target recognition, landmine detection, hyperspectral image analysis, and information fusion.