Machine learning offers a fantastically powerful toolkit for building useful com- plex prediction systems quickly. This paper argues it is dangerous to think of these quick wins as coming for free. Using the software engineering framework of technical debt, we find it is common to incur massive ongoing maintenance costs in real-world ML systems. We explore several ML-specific risk factors to account for in system design. These include boundary erosion, entanglement, hidden feedback loops, undeclared consumers, data dependencies, configuration issues, changes in the external world, and a variety of system-level anti-patterns.
%0 Report
%1 journals/spl/TayL15
%A Sculley, D.
%A Holt, Gary
%A Golovin, Daniel
%A Davydov, Eugene
%A Phillips, Todd
%A Ebner, Dietmar
%A Chaudhary, Vinay
%A Young, Michael
%A Crespo, Jean-Francois
%A Dennison, Dan
%D 2015
%K grundlagen seminar
%T Hidden Technical Debt in Machine Learning Systems
%X Machine learning offers a fantastically powerful toolkit for building useful com- plex prediction systems quickly. This paper argues it is dangerous to think of these quick wins as coming for free. Using the software engineering framework of technical debt, we find it is common to incur massive ongoing maintenance costs in real-world ML systems. We explore several ML-specific risk factors to account for in system design. These include boundary erosion, entanglement, hidden feedback loops, undeclared consumers, data dependencies, configuration issues, changes in the external world, and a variety of system-level anti-patterns.
@techreport{journals/spl/TayL15,
abstract = {Machine learning offers a fantastically powerful toolkit for building useful com- plex prediction systems quickly. This paper argues it is dangerous to think of these quick wins as coming for free. Using the software engineering framework of technical debt, we find it is common to incur massive ongoing maintenance costs in real-world ML systems. We explore several ML-specific risk factors to account for in system design. These include boundary erosion, entanglement, hidden feedback loops, undeclared consumers, data dependencies, configuration issues, changes in the external world, and a variety of system-level anti-patterns.},
added-at = {2023-05-09T08:37:29.000+0200},
author = {Sculley, D. and Holt, Gary and Golovin, Daniel and Davydov, Eugene and Phillips, Todd and Ebner, Dietmar and Chaudhary, Vinay and Young, Michael and Crespo, Jean-Francois and Dennison, Dan},
biburl = {https://puma.ub.uni-stuttgart.de/bibtex/2ce5cea3028caa3a3462e359d23747f84/lucabennardo},
ee = {http://dx.doi.org/10.1109/LSP.2014.2368128},
institution = {Google Inc.},
interhash = {41547ff06384cd0bb5e1e0fd8d8faa29},
intrahash = {ce5cea3028caa3a3462e359d23747f84},
keywords = {grundlagen seminar},
timestamp = {2023-05-09T08:37:29.000+0200},
title = {Hidden Technical Debt in Machine Learning Systems},
year = 2015
}
