Virtual network embedding problem which is NP-hard is a key issue for implementing software-defined network which is brought about by network virtualization. Compared with other studies which focus on designing heuris...Virtual network embedding problem which is NP-hard is a key issue for implementing software-defined network which is brought about by network virtualization. Compared with other studies which focus on designing heuristic algorithms to reduce the hardness of the NP-hard problem we propose a robust VNE algorithm based on component connectivity in large-scale network. We distinguish the different components and embed VN requests onto them respectively. And k-core is applied to identify different VN topologies so that the VN request can be embedded onto its corresponding component. On the other hand, load balancing is also considered in this paper. It could avoid blocked or bottlenecked area of substrate network. Simulation experiments show that compared with other algorithms in large-scale network, acceptance ratio, average revenue and robustness can be obviously improved by our algorithm and average cost can be reduced. It also shows the relationship between the component connectivity including giant component and small components and the performance metrics.展开更多
针对历史数据稀疏性导致推荐算法预测精度低的问题,提出基于多重相似度分析和CatBoost的推荐算法。利用修正的余弦相似度函数求解项目元数据和评分数据的相似矩阵并进行融合;采用大规模信息嵌入网络(large-scale information network em...针对历史数据稀疏性导致推荐算法预测精度低的问题,提出基于多重相似度分析和CatBoost的推荐算法。利用修正的余弦相似度函数求解项目元数据和评分数据的相似矩阵并进行融合;采用大规模信息嵌入网络(large-scale information network embedding,LINE)对融合后的相似矩阵进行多阶相似性分析计算更精确的近邻集;以此作为CatBoost的输入预测项目评分并利用Top-N推荐项目。为验证其有效性,在MovieLens数据集上进行实验并与其它方法对比。实验结果表明,该方法具有更高的推荐精度、更强的稳定性,可解决历史数据稀疏性导致的推荐质量低的问题。展开更多
基金supported in part by the National Natural Science Foundation of China under Grant No.61471055
文摘Virtual network embedding problem which is NP-hard is a key issue for implementing software-defined network which is brought about by network virtualization. Compared with other studies which focus on designing heuristic algorithms to reduce the hardness of the NP-hard problem we propose a robust VNE algorithm based on component connectivity in large-scale network. We distinguish the different components and embed VN requests onto them respectively. And k-core is applied to identify different VN topologies so that the VN request can be embedded onto its corresponding component. On the other hand, load balancing is also considered in this paper. It could avoid blocked or bottlenecked area of substrate network. Simulation experiments show that compared with other algorithms in large-scale network, acceptance ratio, average revenue and robustness can be obviously improved by our algorithm and average cost can be reduced. It also shows the relationship between the component connectivity including giant component and small components and the performance metrics.
文摘针对历史数据稀疏性导致推荐算法预测精度低的问题,提出基于多重相似度分析和CatBoost的推荐算法。利用修正的余弦相似度函数求解项目元数据和评分数据的相似矩阵并进行融合;采用大规模信息嵌入网络(large-scale information network embedding,LINE)对融合后的相似矩阵进行多阶相似性分析计算更精确的近邻集;以此作为CatBoost的输入预测项目评分并利用Top-N推荐项目。为验证其有效性,在MovieLens数据集上进行实验并与其它方法对比。实验结果表明,该方法具有更高的推荐精度、更强的稳定性,可解决历史数据稀疏性导致的推荐质量低的问题。
