目的研究基于样本依赖型的范数正则化学习算法的收敛性分析。方法概率论与数理统计的方法。结果给出了一种用K-泛函表示的收敛速度。结论文中的研究表明,样本依赖型学习算法与通常的核学习算法具有相同的收敛速度。
Aim The convergence for the norm regularized learning algorithm depending upon the sample is analyzed. Methods The probability theory and mathematical statistics are used. Results A kind of convergence rate is provided with a K- functional. Conclusion The results show that the convergence rate of least square regression with sample dependent hypothesis spaces is the same as that of the usual kernel learning scheme.