A Survey of Algorithm Debt in Machine and Deep Learning Systems: Definition, Smells, and Future Work

TL;DR

This survey reviews Algorithm Debt in ML/DL systems, defining its characteristics, identifying its signs, and proposing future research directions to improve system reliability.

Contribution

It expands the definition of Algorithm Debt, uncovers its implicit presence, and identifies its smells based on a review of 42 studies.

Findings

Algorithm Debt impacts ML/DL system performance and scalability.

The review uncovers implicit presence and signs of Algorithm Debt.

Future research directions are highlighted to address Algorithm Debt.

Abstract

The adoption of Machine and Deep Learning (ML/DL) technologies introduces maintenance challenges, leading to Technical Debt (TD). Algorithm Debt (AD) is a TD type that impacts the performance and scalability of ML/DL systems. A review of 42 primary studies expanded AD's definition, uncovered its implicit presence, identified its smells, and highlighted future directions. These findings will guide an AD-focused study, enhancing the reliability of ML/DL systems.