A Light-Weight Visualization Tool for Support Vector Machines

The family of Support Vector Machines (SVM) is one of the most popular supervised learning algorithms today. It is renowned for its outstanding performance and accuracy in many different application domains. In this paper, we present a visualization tool to enhance the utility of SVM. It provides access to the distance measure of each data point to the optimal hyperplane. It also provides the distribution of distance values in the feature space. %The tool also provides the interactive features of panning, zooming and picking, which are useful in inspecting misclassified data points. The tool also incorporates capabilities for panning, zooming, and picking. These interactive features are useful for inspecting misclassified data points to improve the performance of SVM. The tool was used in a cancer detection project using FT-IR spectra. Visualization of the spectra led to the detection and removal of noisy spectra.