Abstract:

In order to solve the non-smooth feature of support vector machine model and improve its classification performance further, a novel smooth function based on the geometry of circle tangent is constructed to smooth the non-differentiable term of support vector machine(SVM), and it proposes a circle tangent smooth support vector machine (CTSSVM). Theoretical analysis proves the global convergence of CTSSVM. Comparing to other smooth approaching functions, its smooth precision has an obvious improvement. Numerical experiments and comparisons demonstrate the efficiency of our scheme is better than other smooth models in terms of classification performance.

Key words:

pattern recognition