Relaxing the Accurate Imputation Assumption in Doubly Robust Learning for Debiased Collaborative Filtering

Haoxuan Li, Chunyuan Zheng, Shuyi Wang, Kunhan Wu, Eric Wang, Peng Wu, Zhi Geng, Xu Chen, Xiao-Hua Zhou
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:29448-29460, 2024.

Abstract

Recommender system aims to recommend items or information that may interest users based on their behaviors and preferences. However, there may be sampling selection bias in the data collection process, i.e., the collected data is not a representative of the target population. Many debiasing methods are developed based on pseudo-labelings. Nevertheless, the validity of these methods relies heavily on accurate pseudo-labelings (i.e., the imputed labels), which is difficult to satisfy in practice. In this paper, we theoretically propose several novel doubly robust estimators that are unbiased when either (a) the pseudo-labelings deviate from the true labels with an arbitrary user-specific inductive bias, item-specific inductive bias, or a combination of both, or (b) the learned propensities are accurate. We further propose a propensity reconstruction learning approach that adaptively updates the constraint weights using an attention mechanism and effectively controls the variance. Extensive experiments show that our approach outperforms the state-of-the-art on one semi-synthetic and three real-world datasets.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-li24cq, title = {Relaxing the Accurate Imputation Assumption in Doubly Robust Learning for Debiased Collaborative Filtering}, author = {Li, Haoxuan and Zheng, Chunyuan and Wang, Shuyi and Wu, Kunhan and Wang, Eric and Wu, Peng and Geng, Zhi and Chen, Xu and Zhou, Xiao-Hua}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {29448--29460}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/li24cq/li24cq.pdf}, url = {https://proceedings.mlr.press/v235/li24cq.html}, abstract = {Recommender system aims to recommend items or information that may interest users based on their behaviors and preferences. However, there may be sampling selection bias in the data collection process, i.e., the collected data is not a representative of the target population. Many debiasing methods are developed based on pseudo-labelings. Nevertheless, the validity of these methods relies heavily on accurate pseudo-labelings (i.e., the imputed labels), which is difficult to satisfy in practice. In this paper, we theoretically propose several novel doubly robust estimators that are unbiased when either (a) the pseudo-labelings deviate from the true labels with an arbitrary user-specific inductive bias, item-specific inductive bias, or a combination of both, or (b) the learned propensities are accurate. We further propose a propensity reconstruction learning approach that adaptively updates the constraint weights using an attention mechanism and effectively controls the variance. Extensive experiments show that our approach outperforms the state-of-the-art on one semi-synthetic and three real-world datasets.} }
Endnote
%0 Conference Paper %T Relaxing the Accurate Imputation Assumption in Doubly Robust Learning for Debiased Collaborative Filtering %A Haoxuan Li %A Chunyuan Zheng %A Shuyi Wang %A Kunhan Wu %A Eric Wang %A Peng Wu %A Zhi Geng %A Xu Chen %A Xiao-Hua Zhou %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-li24cq %I PMLR %P 29448--29460 %U https://proceedings.mlr.press/v235/li24cq.html %V 235 %X Recommender system aims to recommend items or information that may interest users based on their behaviors and preferences. However, there may be sampling selection bias in the data collection process, i.e., the collected data is not a representative of the target population. Many debiasing methods are developed based on pseudo-labelings. Nevertheless, the validity of these methods relies heavily on accurate pseudo-labelings (i.e., the imputed labels), which is difficult to satisfy in practice. In this paper, we theoretically propose several novel doubly robust estimators that are unbiased when either (a) the pseudo-labelings deviate from the true labels with an arbitrary user-specific inductive bias, item-specific inductive bias, or a combination of both, or (b) the learned propensities are accurate. We further propose a propensity reconstruction learning approach that adaptively updates the constraint weights using an attention mechanism and effectively controls the variance. Extensive experiments show that our approach outperforms the state-of-the-art on one semi-synthetic and three real-world datasets.
APA
Li, H., Zheng, C., Wang, S., Wu, K., Wang, E., Wu, P., Geng, Z., Chen, X. & Zhou, X.. (2024). Relaxing the Accurate Imputation Assumption in Doubly Robust Learning for Debiased Collaborative Filtering. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:29448-29460 Available from https://proceedings.mlr.press/v235/li24cq.html.

Related Material