论文信息 - Terminal Replacement Prediction Based on Deep Belief Networks

Terminal Replacement Prediction Based on Deep Belief Networks

To help telecommunications operators accurately predict the terminal replacement behavior, and improve the success rate of marketing and the accuracy of resources devoting, huge user consumption data are used to build Deep Belief Network. The deep features that characterize the terminal replacement behavior are learned, through which a terminal replacement prediction model is conducted. Experiments are carried out on real data set, and the prediction accuracy is over 82%. It is better than three others models based on 1 Nearest Neighbors, Support Vector Machines and Neural Network. The experiments results show that the features obtained by deep learning are more descriptive for predicting terminal replacement behavior.

[1] Yoshua Bengio,et al. Practical Recommendations for Gradient-Based Training of Deep Architectures , 2012, Neural Networks: Tricks of the Trade.

[2] Quoc V. Le,et al. Measuring Invariances in Deep Networks , 2009, NIPS.

[3] Marc'Aurelio Ranzato,et al. Efficient Learning of Sparse Representations with an Energy-Based Model , 2006, NIPS.

[4] Thomas Hofmann,et al. Efficient Learning of Sparse Representations with an Energy-Based Model , 2007 .

[5] Yoshua Bengio,et al. Understanding the difficulty of training deep feedforward neural networks , 2010, AISTATS.

[6] Yoshua Bengio,et al. Why Does Unsupervised Pre-training Help Deep Learning? , 2010, AISTATS.

[7] Yee Whye Teh,et al. A Fast Learning Algorithm for Deep Belief Nets , 2006, Neural Computation.

[8] Yoshua Bengio,et al. Greedy Layer-Wise Training of Deep Networks , 2006, NIPS.

[9] Geoffrey E. Hinton. A Practical Guide to Training Restricted Boltzmann Machines , 2012, Neural Networks: Tricks of the Trade.