[1]聂鑫,孙晓燕,陈杨,等.基于改进Wide&Deep交互特征提取的移动APP转化率预估[J].郑州大学学报(工学版),2020,41(06):26-32.[doi:10.13705/j.issn.1671-6833.2020.06.005]
 Sun Xiaoyan,Nie Xin,Blizzard,et al.Improved Wide&Deep for Interactive Feature Extraction based Mobil APP Conversion Rate Prediction[J].Journal of Zhengzhou University (Engineering Science),2020,41(06):26-32.[doi:10.13705/j.issn.1671-6833.2020.06.005]
点击复制

基于改进Wide&Deep交互特征提取的移动APP转化率预估()
分享到:

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
41卷
期数:
2020年06期
页码:
26-32
栏目:
出版日期:
2020-12-31

文章信息/Info

Title:
Improved Wide&Deep for Interactive Feature Extraction based Mobil APP Conversion Rate Prediction
作者:
聂鑫孙晓燕陈杨暴琳
中国矿业大学信息与控制工程学院,江苏徐州221008, 中国矿业大学信息与控制工程学院,江苏徐州221008, 中国矿业大学信息与控制工程学院,江苏徐州221008, 中国矿业大学信息与控制工程学院,江苏徐州221008

Author(s):
School of Information and Control Engineering, China University of Mining and Technology, 221008, Xuzhou, Jiangsu, School of Information and Control Engineering, China University of Mining and Technology, 221008, Xuzhou, Jiangsu, School of Information and Control Engineering, China University of Mining and Technology, 221008, Xuzhou, Jiangsu, School of Information and Control Engineering, China University 221008

关键词:
Keywords:
DOI:
10.13705/j.issn.1671-6833.2020.06.005
文献标志码:
A
摘要:
移动APP广告转化率预估已成为当前影响广告投放效率、广告排序和收益的关键因素,由于该问题特征维度高而稀疏、特征间高度交互等,使得转化率精准预估面临极大挑战。该文提出一种融合场感知分解机(Field-aware Factorized Machine, FFM)和深度卷积神经网络的改进Wide&Deep模型,以有效获取高维度稀疏特征的低阶和高阶交互关系,从而实现特征自动高效组合,提高移动APP广告转化率预估精度。在给出算法框架基础之上,针对稀疏数据的嵌入,提出了基于宽度模块FFM挖掘低阶特征交互关系的特征组合算法;然后,根据FFM所提取隐特征向量,进一步给出了基于深度模块多层卷积神经网络提取高阶交互关系的特征提取策略;最后,将宽度和深度模块分别获取的特征组合用于转化率预估。所提算法在腾讯移动APP广告转化率预估中的应用表明了该方法在提高预估精度上的有效性。
Abstract:
The prediction of the conversion rate (CRP) of mobile APP advertising is very crucial for the advertisements delivery, ranking and revenue. Due to the high-dimension, sparse, and high interactive, the accurate CRP faces great challenges. In this paper, we propose an improved Wide&Deep model of fusing Field-aware Factorized Machine (FFM) and deep convolutional neural network (DCNN) to effectively and automatically obtain the low-order and high-order interactions of high-dimensional sparse features, so as to realize the automatic and efficient combination of features and improve the accuracy of mobile APP advertising conversion rate estimation. The framework is first delivered, and a feature combination algorithm based on width module FFM to extract the interaction relations of low-order features is presented for the embedding of sparse data. The extraction of the high-order interactive features based on a DCNN is further given by fusing the latent features obtained by the FFM. Finally, the interactive feature combinations obtained by width and depth modules are integrated for the CRP. The application of the proposed algorithm in predicting the conversion rate of Tencent’’’’’’’’s mobile APP advertisements demonstrates the effectiveness of the method in improving the prediction accuracy.
更新日期/Last Update: 2021-02-10