Hidden Technical Debt in Machine Learning Systems

By D. Sculley et al
Read the original document by opening this link in a new tab.

Table of Contents

1. Introduction
2. Complex Models Erode Boundaries
3. Data Dependencies Cost More than Code Dependencies
4. Feedback Loops
5. ML-System Anti-Patterns

Summary

Machine learning offers a powerful toolkit for building complex prediction systems quickly but comes with hidden technical debt. This paper discusses the ongoing maintenance costs incurred in real-world ML systems due to various risk factors like boundary erosion, entanglement, hidden feedback loops, undeclared consumers, and more. It emphasizes the importance of addressing system-level technical debt in ML systems and highlights the challenges in maintaining such systems over time.
×
This is where the content will go.