Generalizing, Optimizing, and Decoding Support Vector Machine Classification

Abstract :

A major challenge in the classification of complex data, that requires the combination of several processing steps, is the selection of the optimal algorithms for preprocessing and classification . Here, we present three steps to face this problem. First, we introduce a generalized model for Support Vector Machine (SVM) variants which generates both unary and online classifiers. This model improves the understanding of relationships between the variants which facilitates the choice and implementation of the classifier. Second, we propose the signal processing and classification environment pySPACE which enables the systematic evaluation and comparison of algorithms. Third, we introduce an approach called backtransformation which enables a visualization of the complete processing chain in the the input data space and thereby allows for a joint interpretation of preprocessing and classification to decode the decision process. Finally, the benefit of combining all three approaches is shown in an application on handwritten digit classification.

Keywords :

pySPACE, support vector machine, relative margin, zero separation approach, online learning, backtransformation