Learning from Delayed Outcomes via Proxies with Applications to Recommender Systems

Timothy Arthur Mann, Sven Gowal, Andras Gyorgy, Huiyi Hu, Ray Jiang, Balaji Lakshminarayanan, Prav Srinivasan
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:4324-4332, 2019.

Abstract

Predicting delayed outcomes is an important problem in recommender systems (e.g., if customers will finish reading an ebook). We formalize the problem as an adversarial, delayed online learning problem and consider how a proxy for the delayed outcome (e.g., if customers read a third of the book in 24 hours) can help minimize regret, even though the proxy is not available when making a prediction. Motivated by our regret analysis, we propose two neural network architectures: Factored Forecaster (FF) which is ideal if the proxy is informative of the outcome in hindsight, and Residual Factored Forecaster (RFF) that is robust to a non-informative proxy. Experiments on two real-world datasets for predicting human behavior show that RFF outperforms both FF and a direct forecaster that does not make use of the proxy. Our results suggest that exploiting proxies by factorization is a promising way to mitigate the impact of long delays in human-behavior prediction tasks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-mann19a, title = {Learning from Delayed Outcomes via Proxies with Applications to Recommender Systems}, author = {Mann, Timothy Arthur and Gowal, Sven and Gyorgy, Andras and Hu, Huiyi and Jiang, Ray and Lakshminarayanan, Balaji and Srinivasan, Prav}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {4324--4332}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/mann19a/mann19a.pdf}, url = {https://proceedings.mlr.press/v97/mann19a.html}, abstract = {Predicting delayed outcomes is an important problem in recommender systems (e.g., if customers will finish reading an ebook). We formalize the problem as an adversarial, delayed online learning problem and consider how a proxy for the delayed outcome (e.g., if customers read a third of the book in 24 hours) can help minimize regret, even though the proxy is not available when making a prediction. Motivated by our regret analysis, we propose two neural network architectures: Factored Forecaster (FF) which is ideal if the proxy is informative of the outcome in hindsight, and Residual Factored Forecaster (RFF) that is robust to a non-informative proxy. Experiments on two real-world datasets for predicting human behavior show that RFF outperforms both FF and a direct forecaster that does not make use of the proxy. Our results suggest that exploiting proxies by factorization is a promising way to mitigate the impact of long delays in human-behavior prediction tasks.} }
Endnote
%0 Conference Paper %T Learning from Delayed Outcomes via Proxies with Applications to Recommender Systems %A Timothy Arthur Mann %A Sven Gowal %A Andras Gyorgy %A Huiyi Hu %A Ray Jiang %A Balaji Lakshminarayanan %A Prav Srinivasan %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-mann19a %I PMLR %P 4324--4332 %U https://proceedings.mlr.press/v97/mann19a.html %V 97 %X Predicting delayed outcomes is an important problem in recommender systems (e.g., if customers will finish reading an ebook). We formalize the problem as an adversarial, delayed online learning problem and consider how a proxy for the delayed outcome (e.g., if customers read a third of the book in 24 hours) can help minimize regret, even though the proxy is not available when making a prediction. Motivated by our regret analysis, we propose two neural network architectures: Factored Forecaster (FF) which is ideal if the proxy is informative of the outcome in hindsight, and Residual Factored Forecaster (RFF) that is robust to a non-informative proxy. Experiments on two real-world datasets for predicting human behavior show that RFF outperforms both FF and a direct forecaster that does not make use of the proxy. Our results suggest that exploiting proxies by factorization is a promising way to mitigate the impact of long delays in human-behavior prediction tasks.
APA
Mann, T.A., Gowal, S., Gyorgy, A., Hu, H., Jiang, R., Lakshminarayanan, B. & Srinivasan, P.. (2019). Learning from Delayed Outcomes via Proxies with Applications to Recommender Systems. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:4324-4332 Available from https://proceedings.mlr.press/v97/mann19a.html.

Related Material