Investigation of kernel functions in EDA-GK

Investigation of kernel functions in EDA-GK
复制标题

DOI:
10.1145/3205651.3208785
复制
发表时间:
2018-07
期刊:
Proceedings of the Genetic and Evolutionary Computation Conference Companion
影响因子:
--
通讯作者:
Ryoichi Hasegawa;H. Handa
Ryoichi Hasegawa;H. Handa
中科院分区:
其他
文献类型:
--
作者:
Ryoichi Hasegawa;H. Handa

文献摘要

相似文献

我们提出了基于图核的分布估计算法EDA-GK。EDA-GK是为解决与图形相关的问题而设计的,其中个体可以用图形来表示。通过使用图的核,EDA-GK可以很好地解决与图相关的问题。EDA-GK使用图核作为EDA中的概率模型。在这项研究中,我们考察了Weisfeler-Lehman核,以及两个核的混合。对图同构问题的实验结果表明了该方法的有效性:
We have proposed EDA-GK, Estimation of Distribution Algorithms with Graph Kernels. The EDA-GK is designed for solving graph-related problems, where individuals can be represented by graphs. By using graph kernels, the EDA-GK can be solved for graph-related problems well. The EDA-GK uses the graph kernels as probabilistic models in EDA. In this study, we examine the Weisfeiler-Lehman Kernel, and the mixture of two kernels. Experimental results on Graph Isomorphism problems showed the effectiveness of the pro- posed method: