核机器集成学习算法的误差分析

Error analysis of learning algorithm of kernel machine ensemble

针对偏置b的一般情况,估计核机器集成学习算法的排一误差和推广误差的界.结果表明,集成学习算法理论上可以提高核机器的稳定性.

Leave-one-out error and generalization error of learning algorithm of kernel machine ensemble with nonzero threshold b are estimated.The results show that in theory the ensemble algorithm can improve stability of the kernel machine.