Large Scale CVR Prediction through Dynamic Transfer Learning of Global and Local Features

Hongxia Yang, Quan Lu, Angus Xianen Qiu, Chun Han
; Proceedings of the 5th International Workshop on Big Data, Streams and Heterogeneous Source Mining: Algorithms, Systems, Programming Models and Applications at KDD 2016, PMLR 53:103-119, 2016.

Abstract

This paper presents a combination of strategies for conversion rate (CVR) prediction de- ployed at the Yahoo! demand side platform (DSP) Brightroll, targeting at modeling extremely high dimensional, sparse data with limited human intervention. We propose a novel probabilistic generative model by tightly integrating components of natural language processing, dynamic transfer learning and scalable prediction, named Dynamic Transfer Learning with Reinforced Word Modeling (a.k.a. Trans-RWM ) to predict user conversion rates. Our model is based on assumptions that: on a higher level, information can be transferable between related campaigns; on a lower level, users who searched similar contents or browsed similar pages would have a higher probability of sharing similar latent purchase interests. Novelties of this framework include (i) A novel natural language modeling specifically tailored for semantic inputs of CVR prediction; (ii) A Bayesian transfer learning model to dynamically transfer the knowledge from source to the future target; (iii) An automatic new updating rule with adaptive regularization using Stochastic Gradient Monte Carlo to support the efficient updating of Trans-RWM in high-dimensional and sparse data. We demonstrate that on Brightroll our framework can effectively discriminate extremely rare events in terms of their conversion propensity.

Cite this Paper


BibTeX
@InProceedings{pmlr-v53-yang16, title = {Large Scale CVR Prediction through Dynamic Transfer Learning of Global and Local Features}, author = {Hongxia Yang and Quan Lu and Angus Xianen Qiu and Chun Han}, booktitle = {Proceedings of the 5th International Workshop on Big Data, Streams and Heterogeneous Source Mining: Algorithms, Systems, Programming Models and Applications at KDD 2016}, pages = {103--119}, year = {2016}, editor = {Wei Fan and Albert Bifet and Jesse Read and Qiang Yang and Philip S. Yu}, volume = {53}, series = {Proceedings of Machine Learning Research}, address = {San Francisco, California, USA}, month = {14 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v53/yang16.pdf}, url = {http://proceedings.mlr.press/v53/yang16.html}, abstract = {This paper presents a combination of strategies for conversion rate (CVR) prediction de- ployed at the Yahoo! demand side platform (DSP) Brightroll, targeting at modeling extremely high dimensional, sparse data with limited human intervention. We propose a novel probabilistic generative model by tightly integrating components of natural language processing, dynamic transfer learning and scalable prediction, named Dynamic Transfer Learning with Reinforced Word Modeling (a.k.a. Trans-RWM ) to predict user conversion rates. Our model is based on assumptions that: on a higher level, information can be transferable between related campaigns; on a lower level, users who searched similar contents or browsed similar pages would have a higher probability of sharing similar latent purchase interests. Novelties of this framework include (i) A novel natural language modeling specifically tailored for semantic inputs of CVR prediction; (ii) A Bayesian transfer learning model to dynamically transfer the knowledge from source to the future target; (iii) An automatic new updating rule with adaptive regularization using Stochastic Gradient Monte Carlo to support the efficient updating of Trans-RWM in high-dimensional and sparse data. We demonstrate that on Brightroll our framework can effectively discriminate extremely rare events in terms of their conversion propensity.} }
Endnote
%0 Conference Paper %T Large Scale CVR Prediction through Dynamic Transfer Learning of Global and Local Features %A Hongxia Yang %A Quan Lu %A Angus Xianen Qiu %A Chun Han %B Proceedings of the 5th International Workshop on Big Data, Streams and Heterogeneous Source Mining: Algorithms, Systems, Programming Models and Applications at KDD 2016 %C Proceedings of Machine Learning Research %D 2016 %E Wei Fan %E Albert Bifet %E Jesse Read %E Qiang Yang %E Philip S. Yu %F pmlr-v53-yang16 %I PMLR %J Proceedings of Machine Learning Research %P 103--119 %U http://proceedings.mlr.press %V 53 %W PMLR %X This paper presents a combination of strategies for conversion rate (CVR) prediction de- ployed at the Yahoo! demand side platform (DSP) Brightroll, targeting at modeling extremely high dimensional, sparse data with limited human intervention. We propose a novel probabilistic generative model by tightly integrating components of natural language processing, dynamic transfer learning and scalable prediction, named Dynamic Transfer Learning with Reinforced Word Modeling (a.k.a. Trans-RWM ) to predict user conversion rates. Our model is based on assumptions that: on a higher level, information can be transferable between related campaigns; on a lower level, users who searched similar contents or browsed similar pages would have a higher probability of sharing similar latent purchase interests. Novelties of this framework include (i) A novel natural language modeling specifically tailored for semantic inputs of CVR prediction; (ii) A Bayesian transfer learning model to dynamically transfer the knowledge from source to the future target; (iii) An automatic new updating rule with adaptive regularization using Stochastic Gradient Monte Carlo to support the efficient updating of Trans-RWM in high-dimensional and sparse data. We demonstrate that on Brightroll our framework can effectively discriminate extremely rare events in terms of their conversion propensity.
RIS
TY - CPAPER TI - Large Scale CVR Prediction through Dynamic Transfer Learning of Global and Local Features AU - Hongxia Yang AU - Quan Lu AU - Angus Xianen Qiu AU - Chun Han BT - Proceedings of the 5th International Workshop on Big Data, Streams and Heterogeneous Source Mining: Algorithms, Systems, Programming Models and Applications at KDD 2016 PY - 2016/12/06 DA - 2016/12/06 ED - Wei Fan ED - Albert Bifet ED - Jesse Read ED - Qiang Yang ED - Philip S. Yu ID - pmlr-v53-yang16 PB - PMLR SP - 103 DP - PMLR EP - 119 L1 - http://proceedings.mlr.press/v53/yang16.pdf UR - http://proceedings.mlr.press/v53/yang16.html AB - This paper presents a combination of strategies for conversion rate (CVR) prediction de- ployed at the Yahoo! demand side platform (DSP) Brightroll, targeting at modeling extremely high dimensional, sparse data with limited human intervention. We propose a novel probabilistic generative model by tightly integrating components of natural language processing, dynamic transfer learning and scalable prediction, named Dynamic Transfer Learning with Reinforced Word Modeling (a.k.a. Trans-RWM ) to predict user conversion rates. Our model is based on assumptions that: on a higher level, information can be transferable between related campaigns; on a lower level, users who searched similar contents or browsed similar pages would have a higher probability of sharing similar latent purchase interests. Novelties of this framework include (i) A novel natural language modeling specifically tailored for semantic inputs of CVR prediction; (ii) A Bayesian transfer learning model to dynamically transfer the knowledge from source to the future target; (iii) An automatic new updating rule with adaptive regularization using Stochastic Gradient Monte Carlo to support the efficient updating of Trans-RWM in high-dimensional and sparse data. We demonstrate that on Brightroll our framework can effectively discriminate extremely rare events in terms of their conversion propensity. ER -
APA
Yang, H., Lu, Q., Xianen Qiu, A. & Han, C.. (2016). Large Scale CVR Prediction through Dynamic Transfer Learning of Global and Local Features. Proceedings of the 5th International Workshop on Big Data, Streams and Heterogeneous Source Mining: Algorithms, Systems, Programming Models and Applications at KDD 2016, in PMLR 53:103-119

Related Material