基于卷积-LSTM网络的广告点击率预测模型研究
|
摘要
点击率预测是计算广告学的核心算法之一。传统浅层模型没有充分考虑到数据之间存在的非线性关系,且使用人工特征提取方法费时费力。针对这些问题,提出了基于卷积(Convolutional Neural Networks)-LSTM(Long Short Term Memory)混合神经网络的广告点击率预测模型。该模型使用卷积神经网络提取高影响力特征,并通过LSTM神经网络的时序性进行预测分类。实验结果证明:与浅层模型或单一结构的神经网络模型相比,基于卷积-LSTM的混合神经网络模型能有效提高广告点击事件的预测准确率。
Click through rate prediction is one of the core algorithms for computational advertising. The traditional shallow model does not fully consider the non-linear relationship between data, and the artificial feature extraction method is time-consuming and laborious. Aiming at these problems, this paper presents a click through rate prediction model based on CNN(Convolutional Neural Networks)-LSTM(Long Short Term Memory)hybrid network. This model uses CNN to extract high-impact features, and predicts and classifies them using the sequence of LSTM. The experimental results show that compared with the shallow model or single neural network model, the hybrid neural network model based on CNNLSTM can effectively improve the accuracy of advertising click through rate prediction.
Click through rate prediction is one of the core algorithms for computational advertising. The traditional shallow model does not fully consider the non-linear relationship between data, and the artificial feature extraction method is time-consuming and laborious. Aiming at these problems, this paper presents a click through rate prediction model based on CNN(Convolutional Neural Networks)-LSTM(Long Short Term Memory)hybrid network. This model uses CNN to extract high-impact features, and predicts and classifies them using the sequence of LSTM. The experimental results show that compared with the shallow model or single neural network model, the hybrid neural network model based on CNNLSTM can effectively improve the accuracy of advertising click through rate prediction.
引文
[1]刘鹏,王超.计算广告:互联网商业变现的市场与技术[M].北京:人民邮电出版社,2015.
[2]周傲英,周敏奇,宫学庆.计算广告:以数据为核心的Web综合应用[J].计算机学报,2011,34(10):1805-1819.
[3]Richardson M,Dominowska E,Ragno R.Predicting clicks:estimating the click-through rate for new ads[C]//International Conference on World Wide Web,2007.
[4]Joachims T.Optimizing search engines using clickthrough data[C]//Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining,2002.
[5]Chapelle O,Zhang Y.A dynamic bayesian network click model for web search ranking[C]//International Conference on World Wide Web,2009.
[6]Rendle S.Factorization machines[C]//IEEE International Conference on Data Mining,2011.
[7]李春红,吴英,覃朝勇.基于LASSO变量选择方法的网络广告点击率预测模型研究[J].数理统计与管理,2016,35(5):803-809.
[8]李思琴.基于深度学习的搜索广告点击率预测方法研究[D].哈尔滨:哈尔滨工业大学,2015.
[9]Zhang Y,Dai H,Xu C,et al.Sequential click prediction for sponsored search with recurrent neural networks[C]//Twenty-Eighth AAAI Conference on Artificial Intelligence,2014.
[10]张志强,周永,谢晓芹,等.基于特征学习的广告点击率预估技术研究[J].计算机学报,2016,39(4):780-794.
[11]周飞燕,金林鹏,董军.卷积神经网络研究综述[J].计算机学报,2017,40(6):1229-1251.
[12]Kim Y.Convolutional neural networks for sentence classification[J].Eprint Arxiv,2014.
[13]Zhou C,Sun C,Liu Z,et al.A C-LSTM neural network for text classification[J].Computer Science,2015,1(4):39-44.
[14]Greff K,Srivastava R K,Koutnik J,et al.LSTM:a search space odyssey[J].IEEE Transactions on Neural Networks&Learning Systems,2017,28(10):2222-2232.
[15]Srivastava N,Hinton G,Krizhevsky A,et al.Dropout:a simple way to prevent neural networks from overfitting[J].Journal of Machine Learning Research,2014,15(1):1929-1958.
[16]Lobo J M,Jiménez Valverde A,Real R.AUC:a misleading measure of the performance of predictive distribution models[J].Global Ecology and Biogeography,2008,17(2):145-151.
[2]周傲英,周敏奇,宫学庆.计算广告:以数据为核心的Web综合应用[J].计算机学报,2011,34(10):1805-1819.
[3]Richardson M,Dominowska E,Ragno R.Predicting clicks:estimating the click-through rate for new ads[C]//International Conference on World Wide Web,2007.
[4]Joachims T.Optimizing search engines using clickthrough data[C]//Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining,2002.
[5]Chapelle O,Zhang Y.A dynamic bayesian network click model for web search ranking[C]//International Conference on World Wide Web,2009.
[6]Rendle S.Factorization machines[C]//IEEE International Conference on Data Mining,2011.
[7]李春红,吴英,覃朝勇.基于LASSO变量选择方法的网络广告点击率预测模型研究[J].数理统计与管理,2016,35(5):803-809.
[8]李思琴.基于深度学习的搜索广告点击率预测方法研究[D].哈尔滨:哈尔滨工业大学,2015.
[9]Zhang Y,Dai H,Xu C,et al.Sequential click prediction for sponsored search with recurrent neural networks[C]//Twenty-Eighth AAAI Conference on Artificial Intelligence,2014.
[10]张志强,周永,谢晓芹,等.基于特征学习的广告点击率预估技术研究[J].计算机学报,2016,39(4):780-794.
[11]周飞燕,金林鹏,董军.卷积神经网络研究综述[J].计算机学报,2017,40(6):1229-1251.
[12]Kim Y.Convolutional neural networks for sentence classification[J].Eprint Arxiv,2014.
[13]Zhou C,Sun C,Liu Z,et al.A C-LSTM neural network for text classification[J].Computer Science,2015,1(4):39-44.
[14]Greff K,Srivastava R K,Koutnik J,et al.LSTM:a search space odyssey[J].IEEE Transactions on Neural Networks&Learning Systems,2017,28(10):2222-2232.
[15]Srivastava N,Hinton G,Krizhevsky A,et al.Dropout:a simple way to prevent neural networks from overfitting[J].Journal of Machine Learning Research,2014,15(1):1929-1958.
[16]Lobo J M,Jiménez Valverde A,Real R.AUC:a misleading measure of the performance of predictive distribution models[J].Global Ecology and Biogeography,2008,17(2):145-151.