arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2005.11018cs.LGcs.ITmath.ITstat.ML
收藏

半监督学习:无标签数据同样有用的情形

Semi-Supervised Learning: the Case When Unlabeled Data is Equally Useful

University of Melbourne(墨尔本大学)

Jingge Zhu

AI总结:

本文研究连续参数统计模型中的半监督学习,证明在特定分布条件下无标签数据可达到与有标签数据同等的学习速率收益。

摘要:

半监督学习算法试图利用成本相对较低的无标签数据来提升学习性能。在本工作中,我们考虑数据分布可由连续参数刻画的统计模型。我们证明,在关于分布的某些条件下,就学习速率而言,无标签数据与有标签数据同样有用。具体而言,设 n、m 分别为有标签数据和无标签数据的数量。结果表明,如果 m∼n,半监督学习的学习速率缩放为 O(1/n);如果对于某个 γ>0 有 m∼n^{1+γ},则缩放为 O(1/n^{1+γ}),而监督学习的学习速率缩放为 O(1/n)。

英文摘要:

Semi-supervised learning algorithms attempt to take advantage of relatively inexpensive unlabeled data to improve learning performance. In this work, we consider statistical models where the data distributions can be characterized by continuous parameters. We show that under certain conditions on the distribution, unlabeled data is equally useful as labeled date in terms of learning rate. Specifically, let $n, m$ be the number of labeled and unlabeled data, respectively. It is shown that the learning rate of semi-supervised learning scales as $O(1/n)$ if $m\sim n$, and scales as $O(1/n^{1+γ})$ if $m\sim n^{1+γ}$ for some $γ>0$, whereas the learning rate of supervised learning scales as $O(1/n)$.

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。

↑