arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.05612cs.LGcs.HCcs.SE
收藏

基于离线强化学习的Web服务个性化

Personalization for Web-based Services using Offline Reinforcement Learning

The University of New Mexico(新墨西哥大学);Facebook Inc.(Facebook公司)

Pavlos Athanasios Apostolopoulos, Zehui Wang, Hanson Wang, Chad Zhou, Kittipat Virochsiri, Norm Zhou, Igor L. Markov

摘要:

大规模的基于Web的服务提供了基于观察到的用户交互来改进UI策略的机会。我们通过无模型的离线强化学习(RL)和离线训练来解决学习这些策略的挑战。部署在一个主要社交网络的用户身份验证生产系统中,它大大提高了长期目标。我们阐述了实际挑战,比较了几种ML技术,提供了关于RL模型的训练和评估的见解,并讨论了泛化。

英文摘要:

Large-scale Web-based services present opportunities for improving UI policies based on observed user interactions. We address challenges of learning such policies through model-free offline Reinforcement Learning (RL) with off-policy training. Deployed in a production system for user authentication in a major social network, it significantly improves long-term objectives. We articulate practical challenges, compare several ML techniques, provide insights on training and evaluation of RL models, and discuss generalizations.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑