Unverified50% confidenceFactExact time
Google's 2017 paper 'Data Management Challenges in Production Machine Learning' points out that the most common technical debt in ML systems comes from data pipelines rather than the models themselves.
1
Sources
50%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
Calibra：机器人学习数据集开源质检工具详解
Related Claims
VerifiedIn 2015, Google systematically elaborated on the technical debt issues of ML systems in the paper 'Hidden Technical Debt in Machine Learning Systems'.79% similarVerifiedIn the paper "Hidden Technical Debt in Machine Learning Systems," Google pointed out that in real ML systems, the actual ML code constitutes only a small fraction of the entire system, surrounded by large amounts of data collection, feature extraction, configuration management, serving infrastructure, and monitoring code.76% similarVerifiedGoogle's 2015 paper "Hidden Technical Debt in Machine Learning Systems" revealed that the actual machine learning code in ML systems accounts for only a very small proportion.71% similarUnverifiedGoogle's machine learning engineers have summarized from experience that actual machine learning code in a machine learning system typically accounts for only 5%, while the remaining 95% consists of data pipelines, service wrappers, and monitoring systems.65% similarUnverified据Google 2023年研究报告，在实际生产ML系统中，模型代码仅占整个系统代码的5-10%62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/775785API
curl https://kongchang.com/api/v1/knowledge/claims/775785MCP
get_claim(id=775785)