| 分布式数据集的稳健统计诊断 |
| 点此下载全文 |
| 引用本文：胡冠浩,姜荣.分布式数据集的稳健统计诊断[J].上海第二工业大学(中文版),2025,42(3):323-329 |
| 摘要点击次数: 379 |
| 全文下载次数: 57 |
|
|
| 中文摘要:随着互联网、物联网、人工智能等领域的飞速发展, 分布式系统的应用场景正在不断拓宽。然而, 由于分布式系统中服务器来源的多样性, 可能存在异质性, 进而影响统计推断的准确性。因此, 在分布式系统中进行统计诊断具有重要意义。采用边际相关性作为诊断统计量, 并借助Huber 回归增强对数据源多样性与重尾噪声影响下的稳健性。数值模拟结果验证了所提方法的有效性, 表明其在处理大规模高维数据集时, 在分布式计算环境中具有良好的适用性与优越性。 |
| 中文关键词:统计诊断 分布式数据 Huber 回归 群组删除 |
| |
| Robust Statistical Diagnostics for Distributed Datasets |
|
|
| Abstract:With the rapid development of fields such as the Internet, the Internet of Things, and Artificial Intelligence, the application scenarios of distributed systems are constantly expanding. However, due to the diversity of server sources in distributed systems, there may be heterogeneity that could affect the accuracy of statistical inference. Therefore, performing statistical diagnostics in distributed systems is of great significance. The marginal correlation is used as a diagnostic statistic and Huber regression is employed to enhance robustness when facing data source diversity and heavy-tailed noise. The results of numerical simulations are used to verify the effectiveness of the proposed method, demonstrating its good applicability and superiority in a distributed computing environment for handling large-scale, high-dimensional datasets. |
| keywords:statistical diagnostics distributed data Huber regression group deletion |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|