Future Gradient Descent for Adapting the Temporal Shifting Data Distribution in Online Recommendation Systems
Abstract
One of the key challenges of learning an online recommendation model is the temporal domain shift, which causes the mismatch between the training and testing data distribution and hence domain generalization error. To overcome, we propose to learn a meta future gradient generator that forecasts the gradient information of the future data distribution for training so that the recommendation model can be trained as if we were able to look ahead at the future of its deployment. Compared with Batch Update, a widely used paradigm, our theory suggests that the proposed algorithm achieves smaller temporal domain generalization error measured by a gradient variation term in a local regret. We demonstrate the empirical advantage by comparing with various representative baselines.
Cite
Text
Ye et al. "Future Gradient Descent for Adapting the Temporal Shifting Data Distribution in Online Recommendation Systems." Uncertainty in Artificial Intelligence, 2022.Markdown
[Ye et al. "Future Gradient Descent for Adapting the Temporal Shifting Data Distribution in Online Recommendation Systems." Uncertainty in Artificial Intelligence, 2022.](https://mlanthology.org/uai/2022/ye2022uai-future/)BibTeX
@inproceedings{ye2022uai-future,
title = {{Future Gradient Descent for Adapting the Temporal Shifting Data Distribution in Online Recommendation Systems}},
author = {Ye, Mao and Jiang, Ruichen and Wang, Haoxiang and Choudhary, Dhruv and Du, Xiaocong and Bhushanam, Bhargav and Mokhtari, Aryan and Kejariwal, Arun and Liu, Qiang},
booktitle = {Uncertainty in Artificial Intelligence},
year = {2022},
pages = {2256-2266},
volume = {180},
url = {https://mlanthology.org/uai/2022/ye2022uai-future/}
}