Classification of Unknown Powders Using a Support Vector Machine Classification Model

• Activity Number: 649
• Type: Topic Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Jessi Cisewski*+ and Jan Hannig and Emily Snyder
• Companies: The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and Environmental Protection Agency
• Keywords: Classification model ; nonparametric statistic ; support vector machine ; wavelet ; dimension reduction

Abstract

When a large building, complex or area has been contaminated with a potentially harmful substance, it can be costly and require an inordinate amount of time to test all unknown substances within the facility to be analyzed in a laboratory by an expert. Portable laser-induced breakdown spectroscopy (LIPS) devices have been developed, which produce spectra of the unknown substances to aid in locating the most contaminated areas. Each spectrum produced must be classified as harmful or not. The proposed method builds a classification model for such spectra relying on the known elemental structure of particular harmful substances. A wavelet transformation is incorporated to allow for possible thresholding or standardization, then a linear model technique using the known elemental structure of the harmful substance is incorporated for dimension reduction, and finally support vector machines are employed for the final classification of the substance. The method has been applied to real-data produced from LIPS devices. Several methods used to test the performance of the classification model reveal promising results.