arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2107.04520cs.LGstat.ML
收藏

标签分布偏移的在线自适应

Online Adaptation to Label Distribution Shift

Cornell University(康奈尔大学);Facebook AI Research(Facebook 人工智能研究院)

Ruihan Wu, Chuan Guo, Yi Su, Kilian Q. Weinberger

摘要:

机器学习模型在现实世界中部署时经常遇到分布偏移。在本文中,我们专注于适应标签分布偏移在线设置,测试时的标签分布是不断变化的,模型必须动态地适应它,而不观察到真正的标签。利用一种新的分析,我们表明,缺乏真正的标签并不妨碍估计的预期测试损失,这使得减少在线标签偏移适应传统的在线学习。根据这一观察,我们提出了经典的在线学习技术,如跟随领导者(FTL)和在线梯度下降(OGD)的启发自适应算法,并推导出他们的遗憾界限。我们根据经验验证我们的研究结果在模拟和现实世界的标签分布偏移,并表明OGD是特别有效和强大的各种具有挑战性的标签偏移的情况。

英文摘要:

Machine learning models often encounter distribution shifts when deployed in the real world. In this paper, we focus on adaptation to label distribution shift in the online setting, where the test-time label distribution is continually changing and the model must dynamically adapt to it without observing the true label. Leveraging a novel analysis, we show that the lack of true label does not hinder estimation of the expected test loss, which enables the reduction of online label shift adaptation to conventional online learning. Informed by this observation, we propose adaptation algorithms inspired by classical online learning techniques such as Follow The Leader (FTL) and Online Gradient Descent (OGD) and derive their regret bounds. We empirically verify our findings under both simulated and real world label distribution shifts and show that OGD is particularly effective and robust to a variety of challenging label shift scenarios.

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。

↑