Robust and Interior Point Optimization Methods in Support Vector Machine Training

点击翻译按钮获取中文摘要

9978813TrafalisThis project will investigate new algorithms for training Support Vector Machines based on Interior Point Methods (IPMs). Training a Support Vector Machine (SVM) requires the solution of a very large quadratic programming (QP) optimization problem. The PI will break this large QP problem into a series of smallest possible QP subproblems by exploiting the special structure of the problem. IPMs will be used for the solution of the resulting QP subproblems. The objective is to handle very large training sets for analyzing massive data. There several classes of IPMs from which to choose to solve the QPsubproblems. The PI will use potential reduction and column generation methods.Investigation of the special structure of the SVM QP problem will be examined by using decomposition techniques of mathematical programming.The proposed project is multidisciplinary, integrating state of the art techniques of mathematical programming and machine learning to solve large scale data mining problems with applications to meteorology, seismology, high energy physics, astronomy, finance and credit card fraud. It will have a significant effect on infrastructure of science and engineering since it presents a new approach for training large scale SVMs.