融合RBF的二值神经网络推荐模型

随着网络通信技术的快速发展和互联网信息资源的大规模扩张,信息过载问题日益严重,传统的信息服务使得这一问题得到缓解。但对具有海量条目的信息,用户要根据自己的喜欢找到想要的目标并不容易。为了解决该问题,提出一种融合径向基函数(RBF)的二值化卷积神经网络的推荐模型。该模型建立在卷积神经网络的基础上将输入数据预处理为0或1,极大节省数据存储空间并提高推荐效率。利用RBF建立可信任的亲属网络,根据亲属网络中的相似用户筛选出有用信息并进行分析做出相应推荐。针对电影推荐进行实验,实验结果表明该方法是有效可行的。

BINARY NEURAL NETWORK RECOMMENDATION MODEL BASED ON RBF

With the rapid development of network communication technology and the great expansion of Internet information resources, information is overloading seriously which is mitigated by the traditional information services. However, it is still not easy for users to find their preferred items among the enormous number items on the Internet. To solve this problem, we proposed a recommendation model using binary convolution neural networks and radical basis function(RBF). Our model based on convolutional neural network preprocessed the input data to 0 or 1, saving data storage space and improving recommendation efficiency. We used RBF to establish a trusted kinship network, and made recommendations based on the useful information screened from similar users in the kinship network. This paper made an experiment on film recommendation. The experimental results show that the method is effective and feasible.