摘要
为解决数据稀疏性问题导致的算法性能问题,文章引入GRNN模型,学习并填充用户评分矩阵,采用协同过滤的方法进行个性化推荐。相较于传统的机器学习方法,GRNN模型具有强大的学习能力以及更少的学习参数。此外,文章提出使用果蝇寻优算法自动寻优平滑因子,在提升算法性能的同时提高预测准确率。实验结果表明,基于果蝇算法优化的GRNN预测模型能显著改善F1及MAE指标。
In order to solve the algorithm performance problem caused by data sparsity problem,this paper introduces GRNN model to learn and populate the user rating matrix for personalized recommendation using collaborative filtering.Compared with traditional machine learning methods,the GRNN model has a powerful learning capability as well as fewer learning parameters.In addition,this paper proposes a fruit fly smoothing algorithm to automatically find the optimal smoothing factor to improve the algorithm performance while increasing the prediction accuracy.The experimental results show that the GRNN prediction model optimized based on the fruit fly algorithm achieves significant improvements in Fi values,MAE indexes.
作者
郭家赫
乔宇
许馨
GUO Jiahe;QIAO Yu;XU Xin(South-Central Minzu University,Wuhan 430074,China)
出处
《计算机应用文摘》
2023年第21期111-114,118,共5页
Chinese Journal of Computer Application
关键词
推荐算法
广义回归神经网络
果蝇优化算法
recommendation algorithm
generalized regression neural network
fruit fly optimization algorithm
