MLINSPECT: A Data Distribution Debugger for Machine Learning Pipelines

MLINSPECT: A Data Distribution Debugger for Machine Learning Pipelines
复制标题

MLINSPECT:用于机器学习管道的数据分布调试器

DOI:
10.1145/3448016.3452759
复制
发表时间:
2021
期刊:
ACM SIGMOD: International Conference on Management of Data
影响因子:
--
通讯作者:
Schelter, Sebastian
Schelter, Sebastian
中科院分区:
--
文献类型:
--
作者:
Grafberger, Stefan;Guha, Shubha;Stoyanovich, Julia;Schelter, Sebastian

文献摘要

参考文献

被引文献

相似文献

机器学习(ML)越来越多地被用于自动化有影响力的决策,而这种广泛使用所带来的风险正在引起政策制定者、科学家和媒体的关注。ML应用程序的输入数据通常非常脆弱,这导致了对其可靠性、可问责性和公平性的担忧。虽然偏见检测不能完全自动化,但计算工具可以帮助确定特定类型的数据问题。我们最近提出了mlinspect,这是一个库,可以对ML预处理管道进行轻量级的基于沿袭的检查。在本演示中,我们将展示如何使用mlininspect检测代表性管道中的数据分布错误。与现有的工作相比,mlinspect对流行的数据科学库(如estimator/transformer管道)的声明性抽象进行操作,可以处理关系和矩阵数据,并且不需要手动代码插装。该图书馆可在https://github.com/stefan-grafberger/mlinspect上公开访问。
Machine Learning (ML) is increasingly used to automate impactful decisions, and the risks arising from this wide-spread use are garnering attention from policymakers, scientists, and the media. ML applications are often very brittle with respect to their input data, which leads to concerns about their reliability, accountability, and fairness. While bias detection cannot be fully automated, computational tools can help pinpoint particular types of data issues.We recently proposed mlinspect, a library that enables lightweight lineage-based inspection of ML preprocessing pipelines. In this demonstration, we show how mlinspect can be used to detect data distribution bugs in a representative pipeline. In contrast to existing work, mlinspect operates on declarative abstractions of popular data science libraries like estimator/transformer pipelines, can handle both relational and matrix data, and does not require manual code instrumentation. The library is publicly available at https://github.com/stefan-grafberger/mlinspect.
FairPrep:在增强公平干预措施的研究中向一流公民推广数据
DOI: 10.5441/002/edbt.2020.41
发表时间: 2020
期刊: 2020
影响因子: --
作者:
Schelter, Sebastian;He, Yuxuan;Khilnani, Jatin;Stoyanovich, Julia
通讯作者: Stoyanovich, Julia
克服机器学习管道中的技术偏见
DOI: --
发表时间: 2020
期刊: Bulletin of the Technical Committee on Data Engineering
影响因子: --
作者:
Schelter, Sebastian;Stoyanovich, Julia
通讯作者: Stoyanovich, Julia
模型数据库
DOI: --
发表时间: 2014
期刊: Encyclopedia of Computational Neuroscience
影响因子: --
作者:
R. McDougal;Rixin Wang;Thomas M. Morse;M. Migliore;Luis N. Marenco;N. Carnevale;Michael L. Hines;Gordon M. Shepherd
通讯作者: Gordon M. Shepherd
本机机器学习管道中数据预处理的轻量级检查
DOI: --
发表时间: 2021
期刊: Conference on Innovative Data Systems Research (CIDR
影响因子: --
作者:
Grafberger, Stefan;Stoyanovich, Julia;Schelter, Sebastian
通讯作者: Schelter, Sebastian