arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1703.07596stat.ML
收藏

对称噪声不变性分布的检验与学习

Testing and Learning on Distributions with Symmetric Noise Invariance

Ho Chung Leon Law, Christopher Yau, Dino Sejdinovic

摘要:

分布的核嵌入和最大平均离散度(MMD),分布之间的距离,是完全非参数双样本测试和学习分布的有用工具。然而,样本之间所有可能的差异很少引起关注-发现的差异可能是由于不同类型的测量噪声、数据收集伪影或其他不相关的可变性来源。我们提出了分布之间的距离编码不变性加性对称噪声,旨在测试是否假设真正的底层过程不同。此外,我们构造的分布不变的功能,导致学习算法的鲁棒性与对称加性噪声的输入分布的损害。

英文摘要:

Kernel embeddings of distributions and the Maximum Mean Discrepancy (MMD), the resulting distance between distributions, are useful tools for fully nonparametric two-sample testing and learning on distributions. However, it is rarely that all possible differences between samples are of interest -- discovered differences can be due to different types of measurement noise, data collection artefacts or other irrelevant sources of variability. We propose distances between distributions which encode invariance to additive symmetric noise, aimed at testing whether the assumed true underlying processes differ. Moreover, we construct invariant features of distributions, leading to learning algorithms robust to the impairment of the input distributions with symmetric additive noise.

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。

↑