CHARACTERIZING THE SOLUTION PATH OF MULTICATEGORY SUPPORT VECTOR MACHINES
Yoonkyung Lee, Zhenhuan Cui · 2006
An algorithm for tting the entire regularization path of the support vector machine (SVM) was recently proposed by Hastie et al. (2004). It allows eectiv e computation of solutions and greatly facilitates the choice of the regular- ization parameter that balances a trade-o between complexity of a solution and its t to data. Extending the idea to more general setting of the multiclass case, we characterize the coecien t path of the multicategory SVM via the complemen- tarity conditions for optimality. The extended algorithm provides a computational shortcut to attain the entire spectrum of solutions from the most regularized to the completely overtted ones.