Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2023
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2301.08837 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916833063337984 |
|---|---|
| author | Dwork, Cynthia Lee, Daniel Lin, Huijia Tankala, Pranay |
| author_facet | Dwork, Cynthia Lee, Daniel Lin, Huijia Tankala, Pranay |
| contents | We identify and explore connections between the recent literature on multi-group fairness for prediction algorithms and the pseudorandomness notions of leakage-resilience and graph regularity. We frame our investigation using new variants of multicalibration based on statistical distance and closely related to the concept of outcome indistinguishability. Adopting this perspective leads us not only to new, more efficient algorithms for multicalibration, but also to our graph theoretic results and a proof of a novel hardcore lemma for real-valued functions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2301_08837 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | From Pseudorandomness to Multi-Group Fairness and Back Dwork, Cynthia Lee, Daniel Lin, Huijia Tankala, Pranay Machine Learning Computational Complexity We identify and explore connections between the recent literature on multi-group fairness for prediction algorithms and the pseudorandomness notions of leakage-resilience and graph regularity. We frame our investigation using new variants of multicalibration based on statistical distance and closely related to the concept of outcome indistinguishability. Adopting this perspective leads us not only to new, more efficient algorithms for multicalibration, but also to our graph theoretic results and a proof of a novel hardcore lemma for real-valued functions. |
| title | From Pseudorandomness to Multi-Group Fairness and Back |
| topic | Machine Learning Computational Complexity |
| url | https://arxiv.org/abs/2301.08837 |