Using Balancing Terms to Avoid Discrimination in Classification

Enni, Simon; Assent, Ira

doi:10.1109/icdm.2018.00116

Cited by 1 publication

(1 citation statement)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is worth noting that the optimization objective can account for more things than just predictive power, e.g., classi ers can be optimized for increased fairness [11], or they can be optimized to shi a ention to pa erns considered more sensible [38]. If important goals or constraints are not accounted for in the optimization objective, the learner might output a model that is inconsistent with those.…”

Section: Optimization Objectivementioning

confidence: 99%

Learning by Design: Structuring and Documenting the Human Choices in Machine Learning Development

Enni,

Assent

2021

Preprint

Self Cite

View full text Add to dashboard Cite

e in uence of machine learning (ML) is quickly spreading, and a number of recent technological innovations have applied ML as a central technology. However, ML development still requires a substantial amount of human expertise to be successful. e deliberation and expert judgment applied during ML development cannot be revisited or scrutinized if not properly documented, and this hinders the further adoption of ML technologies-especially in safety critical situations. In this paper, we present a method consisting of eight design questions, that outline the deliberation and normative choices going into creating a ML model. Our method a ords several bene ts, such as supporting critical assessment through methodological transparency, aiding in model debugging, and anchoring model explanations by commi ing to a pre hoc expectation of the model's behavior. We believe that our method can help ML practitioners structure and justify their choices and assumptions when developing ML models, and that it can help bridge a gap between those inside and outside the ML eld in understanding how and why ML models are designed and developed the way they are.

show abstract

Section: Optimization Objectivementioning

confidence: 99%