基于组合核和调和函数的半监督学习
首发时间:2012-02-13
摘要:提出一种基于组合核和调和函数的半监督学习算法。首先构建多个不同结构的图,每个图对应一个图核;然后寻找这些图核的凸优化组合,计算得到一个优化组合核;计算这个组合核对应的图结构,利用得到的优化组合图,结合调和函数思想进行回归分析。实验结果表明所提算法具有良好的回归估计性能,学习精度较高。
For information in English, please click here
Semi-Supervised Learning Based on Combination-Kernel and Harmonic Function
Abstract:A kind of semi-supervised learning based on combination-kernel and harmonic function was proposed in the paper. At first, multiple graphs with different structures were constructed. Each graph corresponds with each graph kernel. Then an optimized combination-kernel was computed through carrying out a convex optimization operation on these basic graph kernels. In the third step, an optimal combination-graph corresponding with the combination-kernel was obtained. At last, based the obtained combination-graph and the idea of harmonic function, regression analysis was carried out. Experimental results show that the proposed regression algorithm has perfect regression property and has high learning precision.
Keywords: Combination-kernel Harmonic function Semi-supervised learning
论文图表:
引用
No.4461667586478132****
同行评议
共计0人参与
勘误表
基于组合核和调和函数的半监督学习
评论
全部评论0/1000