research-article

Multiple robust learning for recommendation

AUTHORs:

Peng WuAuthors Info & Claims

AAAI'23/IAAI'23/EAAI'23: Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelligence and Thirty-Fifth Conference on Innovative Applications of Artificial Intelligence and Thirteenth Symposium on Educational Advances in Artificial Intelligence

Article No.: 493, Pages 4417 - 4425

https://doi.org/10.1609/aaai.v37i4.25562

Published: 07 February 2023 Publication History

Abstract

In recommender systems, a common problem is the presence of various biases in the collected data, which deteriorates the generalization ability of the recommendation models and leads to inaccurate predictions. Doubly robust (DR) learning has been studied in many tasks in RS, with the advantage that unbiased learning can be achieved when either a single imputation or a single propensity model is accurate. In this paper, we propose a multiple robust (MR) estimator that can take the advantage of multiple candidate imputation and propensity models to achieve unbiasedness. Specifically, the MR estimator is unbiased when any of the imputation or propensity models, or a linear combination of these models is accurate. Theoretical analysis shows that the proposed MR is an enhanced version of DR when only having a single imputation and a single propensity model, leading to a smaller bias. Inspired by the derived generalization error bound of MR, we further propose a novel multiple robust learning approach for stabilization. We conduct extensive experiments on real-world and semi-synthetic datasets, which demonstrates the superiority of the proposed approach over the state-of-the-art methods.

References

[1]

Bonner, S.; and Vasile, F. 2018. Causal embeddings for recommendation. In RecSys.

[2]

Chan, K. C. G. 2013. A simple multiple robust estimator for missing response problem. Stat, 2: 143-149.

[3]

Chan, K. C. G.; and Yam, S. C. P. 2014. Oracle, Multiple Robust and Multipurpose Calibration in a Missing Response Problem. Statistical Science, 29: 380-396.

[4]

Chen, J.; Dong, H.; Qiu, Y.; He, X.; Xin, X.; Chen, L.; Lin, G.; and Yang, K. 2021. AutoDebias: Learning to Debias for Recommendation. In SIGIR.

[5]

Chen, J.; Dong, H.; Wang, X.; Feng, F.; Wang, M.; and He, X. 2020. Bias and Debias in Recommender System: A Survey and Future Directions. arXiv:2010.03240.

[6]

Chen, S.; and Haziza, D. 2017. Multiply robust imputation procedures for the treatment of item nonresponse in surveys. Biometrika, 104: 439-453.

[7]

Dai, Q.; Li, H.; Wu, P.; Dong, Z.; Zhou, X.-H.; Zhang, R.; He, X.; Zhang, R.; and Sun, J. 2022. A Generalized Doubly Robust Learning Framework for Debiasing Post-Click Conversion Rate Prediction. In KDD.

[8]

Ding, S.; Wu, P.; Feng, F.; He, X.; Wang, Y.; Liao, Y.; and Zhang, Y. 2022. Addressing Unmeasured Confounder for Recommendation with Sensitivity Analysis. In KDD.

[9]

Duan, X.; and Yin, G. 2017. Ensemble Approaches to Estimating the Population Mean with Missing Response. Scandinavian Journal of Statistics, 44: 899-917.

[10]

Guo, S.; Zou, L.; Liu, Y.; Ye, W.; Cheng, S.; Wang, S.; Chen, H.; Yin, D.; and Chang, Y. 2021. Enhanced Doubly Robust Learning for Debiasing Post-Click Conversion Rate Estimation. In SIGIR.

[11]

Han, P. 2016. Combining Inverse Probability Weighting and Multiple Imputation to Improve Robustness of Estimation. Scandinavian Journal of Statistics, 43: 246-260.

[12]

Han, P. 2018. Calibration and Multiple Robustness When Data Are Missing Not At Random. Statistical Science, 28: 1725-1740.

[13]

Han, P.; and Wang, L. 2013. Estimation with missing data: beyond double robustness. Biometrika, 100: 417-430.

[14]

He, X.; Liao, L.; Zhang, H.; Nie, L.; Hu, X.; and Chua, T. 2017. Neural Collaborative Filtering. In Barrett, R.; Cummings, R.; Agichtein, E.; and Gabrilovich, E., eds., WWW.

[15]

Jiang, N.; and Li, L. 2016. Doubly Robust Off-policy Value Evaluation for Reinforcement Learning. In Balcan, M.; and Weinberger, K. Q., eds., ICML.

[16]

Kingma, D. P.; and Ba, J. 2015. Adam: A Method for Stochastic Optimization. In Bengio, Y.; and LeCun, Y., eds., ICLR.

[17]

Kiyohara, H.; Saito, Y.; Matsuhiro, T.; Narita, Y.; Shimizu, N.; and Yamamoto, Y. 2022. Doubly Robust Off-Policy Evaluation for Ranking Policies under the Cascade Behavior Model. In WSDM.

[18]

Koren, Y.; Bell, R. M.; and Volinsky, C. 2009. Matrix Factorization Techniques for Recommender Systems. Computer, 42(8): 30-37.

Digital Library

[19]

Li, H.; Lyu, Y.; Zheng, C.; and Wu, P. 2023a. TDR-CL: Targeted Doubly Robust Collaborative Learning for Debiased Recommendations. In ICLR.

[20]

Li, H.; Xiao, Y.; Zheng, C.; and Wu, P. 2023b. Balancing Unobserved Confounding with a Few Unbiased Ratings in Debiased Recommendations. In WWW.

[21]

Li, H.; Zheng, C.; and Wu, P. 2023. StableDR: Stabilized Doubly Robust Learning for Recommendation on Data Missing Not at Random. In ICLR.

[22]

Li, W.; Gu, Y.; and Liu, L. 2020. Demystifying a class of multiple robust estimators. Biometrika, 107: 919-933.

[23]

Liu, D.; Cheng, P.; Dong, Z.; He, X.; Pan, W.; and Ming, Z. 2020. A general knowledge distillation framework for counterfactual recommendation via uniform data. In SIGIR.

[24]

Liu, D.; Cheng, P.; Zhu, H.; Dong, Z.; He, X.; Pan, W.; and Ming, Z. 2021. Mitigating Confounding Bias in Recommendation via Information Bottleneck. In RecSys.

[25]

Marlin, B.; Zemel, R. S.; Roweis, S.; and Slaney, M. 2012. Collaborative filtering and the missing at random assumption. arXiv:1206.5267.

[26]

Marlin, B. M.; and Zemel, R. S. 2009. Collaborative prediction and ranking with non-random missing data. In RecSys.

[27]

Paszke, A.; Gross, S.; Massa, F.; Lerer, A.; Bradbury, J.; Chanan, G.; Killeen, T.; Lin, Z.; Gimelshein, N.; Antiga, L.; Desmaison, A.; Köpf, A.; Yang, E. Z.; DeVito, Z.; Raison, M.; Tejani, A.; Chilamkurthy, S.; Steiner, B.; Fang, L.; Bai, J.; and Chintala, S. 2019. PyTorch: An Imperative Style, High-Performance Deep Learning Library. In NeurIPS.

[28]

Saito, Y. 2019. Unbiased Pairwise Learning from Implicit Feedback. In NeurIPS Workshop.

[29]

Saito, Y. 2020. Doubly robust estimator for ranking metrics with post-click conversions. In RecSys.

[30]

Saito, Y.; Yaginuma, S.; Nishino, Y.; Sakata, H.; and Nakata, K. 2020. Unbiased Recommender Learning from Missing-Not-At-Random Implicit Feedback. In Caverlee, J.; Hu, X. B.; Lalmas, M.; and Wang, W., eds., WSDM.

[31]

Sato, M.; Singh, J.; Takemori, S.; Sonoda, T.; Zhang, Q.; and Ohkuma, T. 2019. Uplif-based Evaluation and Optimization of Recommenders. In RecSys.

[32]

Sato, M.; Takemori, S.; Singh, J.; and Ohkuma, T. 2020. Unbiased Learning for the Causal Effect of Recommendation. In RecSys.

[33]

Schnabel, T.; Swaminathan, A.; Singh, A.; Chandak, N.; and Joachims, T. 2016. Recommendations as Treatments: Debiasing Learning and Evaluation. In ICML.

[34]

Steck, H. 2010. Training and testing of recommender systems on data missing not at random. In KDD.

[35]

Steck, H. 2013. Evaluation of recommendations: rating-prediction and ranking. In RecSys.

[36]

Swaminathan, A.; and Joachims, T. 2015. The Self-Normalized Estimator for Counterfactual Learning. In Cortes, C.; Lawrence, N. D.; Lee, D. D.; Sugiyama, M.; and Garnett, R., eds., NeurIPS.

[37]

Tan, Z. 2007. Comment: understanding OR, PS and DR. Statistical Science, 22: 560-568.

[38]

Wang, X.; Zhang, R.; Sun, Y.; and Qi, J. 2019. Doubly Robust Joint Learning for Recommendation on Data Missing Not at Random. In ICML.

[39]

Wang, X.; Zhang, R.; Sun, Y.; and Qi, J. 2021. Combating Selection Biases in Recommender Systems with A Few Unbiased Ratings. In WSDM.

[40]

Wang, Z.; Chen, X.; Wen, R.; Huang, S.-L.; Kuruoglu, E. E.; and Zheng, Y. 2020. Information Theoretic Counterfactual Learning from Missing-Not-At-Random Feedback. NeurIPS.

[41]

Wu, P.; Han, S.; Tong, X.; and Li, R. 2022a. Propensity score regression for causal inference with treatment heterogeneity. Statistica Sinica.

[42]

Wu, P.; Li, H.; Deng, Y.; Hu, W.; Dai, Q.; Dong, Z.; Sun, J.; Zhang, R.; and Zhou, X.-H. 2022b. On the Opportunity of Causal Learning in Recommendation Systems: Foundation, Estimation, Prediction and Challenges. In IJCAI.

[43]

Yang, L.; Cui, Y.; Xuan, Y.; Wang, C.; Belongie, S.; and Estrin, D. 2018. Unbiased offline recommender evaluation for missing-not-at-random implicit feedback. In RecSys.

[44]

Yuan, B.; Hsia, J.; Yang, M.; Zhu, H.; Chang, C.; Dong, Z.; and Lin, C. 2019. Improving Ad Click Prediction by Considering Non-displayed Events. In CIKM.

[45]

Zhang, W.; Bao, W.; Liu, X.-Y.; Yang, K.; Lin, Q.; Wen, H.; and Ramezani, R. 2020. Large-scale Causal Approaches to Debiasing Post-click Conversion Rate Estimation with Multi-task Learning. In WWW.

Cited By

Su HMeng LZhu LLu KLi JHui Yang GWang HHan SHauff CZuccon GZhang Y(2024)DDPO: Direct Dual Propensity Optimization for Post-Click Conversion Rate EstimationProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval10.1145/3626772.3657817(1179-1188)Online publication date: 10-Jul-2024
https://dl.acm.org/doi/10.1145/3626772.3657817
Wang LMa CWu XQiu ZZheng YChen XChua TNgo CKa-Wei Lee RKumar RLauw H(2024)Causally Debiased Time-aware RecommendationProceedings of the ACM Web Conference 202410.1145/3589334.3645400(3331-3342)Online publication date: 13-May-2024
https://dl.acm.org/doi/10.1145/3589334.3645400
Zhang XHuang CZheng KSu HJi TWang WQi HLi JChua TNgo CKa-Wei Lee RKumar RLauw H(2024)Adversarial-Enhanced Causal Multi-Task Framework for Debiasing Post-Click Conversion Rate EstimationProceedings of the ACM Web Conference 202410.1145/3589334.3645379(3287-3296)Online publication date: 13-May-2024
https://dl.acm.org/doi/10.1145/3589334.3645379
Show More Cited By

Recommendations

Bias in Robust Estimation Caused by Discontinuities and Multiple Structures

When fitting models to data containing multiple structures, such as when fitting surface patches to data taken from a neighborhood that includes a range discontinuity, robust estimators must tolerate both gross outliers and pseudo outliers. Pseudo ...
Robust variance estimation for random effects meta-analysis

In random effects meta-analysis, an overall effect is estimated using a weighted mean, with weights based on estimated marginal variances. The variance of the overall effect is often estimated using the inverse of the sum of the estimated weights, and ...
Distributionally Robust Learning

This monograph develops a comprehensive statistical learning framework that is robust to (distributional) perturbations in the data using Distributionally Robust Optimization (DRO) under the Wasserstein metric. Beginning with fundamental properties of the ...

Comments

Information & Contributors

Information

Published In

cover image Guide Proceedings

AAAI'23/IAAI'23/EAAI'23: Proceedings of the Thirty-Seventh AAAI Conference on Artificial Intelligence and Thirty-Fifth Conference on Innovative Applications of Artificial Intelligence and Thirteenth Symposium on Educational Advances in Artificial Intelligence

February 2023

16496 pages

ISBN:978-1-57735-880-0

Copyright © 2023 Association for the Advancement of Artificial Intelligence.

Sponsors

Association for the Advancement of Artificial Intelligence

Publisher

AAAI Press

Publication History

Published: 07 February 2023

Qualifiers

Research-article
Research
Refereed limited

Contributors

Other Metrics

View Article Metrics

Bibliometrics & Citations

Bibliometrics

Article Metrics

7
Total Citations
View Citations
0
Total Downloads

Downloads (Last 12 months)0
Downloads (Last 6 weeks)0

Reflects downloads up to 10 Oct 2024

Other Metrics

View Author Metrics

Citations

Cited By

Su HMeng LZhu LLu KLi JHui Yang GWang HHan SHauff CZuccon GZhang Y(2024)DDPO: Direct Dual Propensity Optimization for Post-Click Conversion Rate EstimationProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval10.1145/3626772.3657817(1179-1188)Online publication date: 10-Jul-2024
https://dl.acm.org/doi/10.1145/3626772.3657817
Wang LMa CWu XQiu ZZheng YChen XChua TNgo CKa-Wei Lee RKumar RLauw H(2024)Causally Debiased Time-aware RecommendationProceedings of the ACM Web Conference 202410.1145/3589334.3645400(3331-3342)Online publication date: 13-May-2024
https://dl.acm.org/doi/10.1145/3589334.3645400
Zhang XHuang CZheng KSu HJi TWang WQi HLi JChua TNgo CKa-Wei Lee RKumar RLauw H(2024)Adversarial-Enhanced Causal Multi-Task Framework for Debiasing Post-Click Conversion Rate EstimationProceedings of the ACM Web Conference 202410.1145/3589334.3645379(3287-3296)Online publication date: 13-May-2024
https://dl.acm.org/doi/10.1145/3589334.3645379
Li HWu KZheng CXiao YWang HGeng ZFeng FHe XWu POh ANaumann TGloberson ASaenko KHardt MLevine S(2023)Removing hidden confounding in recommendationProceedings of the 37th International Conference on Neural Information Processing Systems10.5555/3666122.3668502(54614-54626)Online publication date: 10-Dec-2023
https://dl.acm.org/doi/10.5555/3666122.3668502
Li HXiao YZheng CWu PCui PKrause ABrunskill ECho KEngelhardt BSabato SScarlett J(2023)Propensity mattersProceedings of the 40th International Conference on Machine Learning10.5555/3618408.3619239(20182-20194)Online publication date: 23-Jul-2023
https://dl.acm.org/doi/10.5555/3618408.3619239
Huang SLi HLi QZheng CLiu LEl Saddik AMei TCucchiara RBertini MTobon Vallejo DAtrey PHossain M(2023)Pareto Invariant Representation Learning for Multimedia RecommendationProceedings of the 31st ACM International Conference on Multimedia10.1145/3581783.3612591(6410-6419)Online publication date: 26-Oct-2023
https://dl.acm.org/doi/10.1145/3581783.3612591
Zhang ZDai QChen XDong ZTang RChen HDuh WHuang HKato MMothe JPoblete B(2023)Robust Causal Inference for Recommender System to Overcome Noisy ConfoundersProceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval10.1145/3539618.3592055(2349-2353)Online publication date: 19-Jul-2023
https://dl.acm.org/doi/10.1145/3539618.3592055

View Options

View options

Media

Figures

Other

Tables

View Table of Contents