Cascaded SVMs in Pattern Classification for Time-Sensitive Separating

This paper introduced some fundamental background knowledge of Support Vector Machine, including VC dimension, separable hyperplane, and feature space as well. Two kinds of cascaded SVMs architecture are reviewed, i.e. parallel and serial structure Parallel SVMs can reduce the run time effectively. And serial SVMs is being used for multi-class separating, and discard the negative samples at the early stage. In the end we proposed two potential applications using cascaded SVMs.