A DC Algorithm Scheme For Generalized Eigenvalue Classifiers

Generalized eigenvalue classffiers handle the classification problem very well. However, the major disadvantages of generalized eigenvalue based method lie in that they may be time-consuming and fail to get a stable or optimal solution due to the matrix singularity. This paper proposes a new method using a DC Algorithm scheme to solve a fractional optimization models which ensure the optimality of the solution. Numerical experiments show that the proposed method can get better results for public data sets.