Fair Neighbor Embedding

Tutkimustuotos: KonferenssiartikkeliScientificvertaisarvioitu

1 Lataukset (Pure)

Abstrakti

We consider fairness in dimensionality reduction (DR). Nonlinear DR yields low dimensional representations that let users visualize and explore high-dimensional data. However, traditional DR may yield biased visualizations overemphasizing relationships of societal phenomena to sensitive attributes or protected groups. We introduce a framework of fair neighbor embedding, the Fair Neighbor Retrieval Visualizer, formulating fair nonlinear DR as an information retrieval task with performance and fairness quantified by information retrieval criteria. The method optimizes low-dimensional embeddings that preserve high-dimensional data neighborhoods without biased association of such neighborhoods to protected groups. In experiments the method yields fair visualizations outperforming previous methods.

AlkuperäiskieliEnglanti
OtsikkoProceedings of the 40th International Conference on Machine Learning
Sivut27564-27584
Sivumäärä21
Vuosikerta202
TilaJulkaistu - heinäk. 2023
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisussa
TapahtumaInternational Conference on Machine Learning - Honolulu, Hawaii, Yhdysvallat
Kesto: 23 heinäk. 202329 heinäk. 2023

Julkaisusarja

NimiProceedings of Machine Learning Research
ISSN (painettu)2640-3498

Conference

ConferenceInternational Conference on Machine Learning
Maa/AlueYhdysvallat
KaupunkiHonolulu, Hawaii
Ajanjakso23/07/2329/07/23

Julkaisufoorumi-taso

  • Jufo-taso 3

!!ASJC Scopus subject areas

  • Artificial Intelligence
  • Software
  • Control and Systems Engineering
  • Statistics and Probability

Sormenjälki

Sukella tutkimusaiheisiin 'Fair Neighbor Embedding'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

Siteeraa tätä