Nowadays, many crucial algorithms (recommendation, scoring, ranking and classification) are operated at third party providers, without users or institutions having any insights on how they operate on their data. Such a black-box setup challenges scientists to show what is feasible in terms of audits, both on theoretical and efficiency-oriented aspects.
This workshop aims at summing up the current state of algorithmic-audits through recent scientific advances.
The workshop will take place online (Zoom), please register to attend (see below).
14:45-15:00(Paris)/08:45-09:00(New York): Welcome
15:00-15:30(Paris)/09:00-09:30(NY): Tubes & Bubbles - Topological confinement of YouTube recommendations (PLOS ONE 2020)
15:30-16:00(Paris)/09:30-10:00(NY): On the relevance of APIs facing fairwashed audits (arXiv 2023)
16:00-16:30(Paris)/10:00-10:30(NY): Confidential-PROFITT: Confidential PROof of FaIr Training of Trees (ICLR 2023)
16:30-17:00(Paris)/10:30-11:00(NY): Auditing for discrimination in ad delivery, with and without platform support (CSCW 2023)
17:00-17:30(Paris)/11:00-11:30(NY): A zest of lime: towards architecture-independent model distances (ICLR 2022)
17:30-18:00(Paris)/11:30-12:00(NY): Active fairness auditing (ICML 2022)
18:00(Paris)/12:00(NY) : wrap-up
A poster for publicizing the event.
Timezones: