What question did this study set out to answer?

Establish a comprehensive conceptual framework to identify, analyze, and categorize the specific origins of downstream harm throughout the machine learning life cycle.

October 5, 2021Open Access

A Framework for Understanding Sources of Harm throughout the Machine Learning Life Cycle

View Full Paper

Ask AI

Bookmark

Authors

HSHarini SureshJohn Brown University JGJohn V. GuttagCross-Cutting Cardiology

Discussion

Member takes

Implication

Conceptual framework identifies seven distinct sources of harm across machine learning pipelines, highlighting targeted strategies to anticipate and mitigate algorithmic bias.

Key Points

Establish a comprehensive conceptual framework to identify, analyze, and categorize the specific origins of downstream harm throughout the machine learning life cycle.
Analyzed failure modes across three primary phases of the machine learning pipeline: data collection, model development, and deployment.
Categorized mechanisms of unintended negative outcomes into seven distinct sources of harm.
Identified seven unique operational stages where bias, misalignments, or systemic harms can be introduced into machine learning systems.
Established a structured taxonomy designed to improve diagnostic communication and guide targeted, context-specific mitigation strategies.

Cite This Study

Suresh et al. (2021) studied this question.

synapsesocial.com/papers/6a0a0dfdb0d552aa8b4603f2 https://doi.org/10.1145/3465416.3483305

View Full Paper

Ask AI

Bookmark

Implication

Conceptual framework identifies seven distinct sources of harm across machine learning pipelines, highlighting targeted strategies to anticipate and mitigate algorithmic bias.

A Framework for Understanding Sources of Harm throughout the Machine Learning Life Cycle

Discussion

Member takes

Implication

Key Points

Cite This Study

A Framework for Understanding Sources of Harm throughout the Machine Learning Life Cycle

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider