Professional Documents
Culture Documents
Vanguard Projects: ML Fairness
Vanguard Projects: ML Fairness
The
ML Vanguard Projects
Fairness
Project These projects have committed to making fairness a P1,
going above and beyond to ensure Google's products work
Home for everyone.
About
Vanguard Projects
Infrastructure Infrastructure
Resources
Scorecard Fairness Policy
Presentations
Establish proactive-oriented best-practices and Establish the company-wide normative and
Reports the formal process framework to support fair value-based positions necessary for teams to
Community and inclusive ML development; lowering risk, pursue fairness work. These company-wide
opening up opportunities for market growth, policies will be complimentary to the Scorecard,
FAQ and making remediation easier, should and include positions on risk identi cation,
Archive unfairness arise. sensitive subgroups, and incident response.
Project POC: Andrew Zaldivar Project POC: Jakara Mato
GET INVOLVED
Infrastructure Production
Production
Production
YouTube Restricted Mode
Display Ads Quality Signals
Ameliorate unnecessary ltering of LGBTQ
Correct or mitigate unfair inferences used in content in Restricted Mode and expand work to
Display Ads targeting. other YouTube classi ers.
Project POCs: Marzia Polito and Jen Io nova Project POC: Tulsee Doshi
Production Production
Production
https://ml-fairness.googleplex.com/vanguard.html 1/3
8/17/2017 ML Fairness
Production
Production
Conversation-AI
Links
Mitigating false positives for non-toxic
comments that contain language related to MomentQuality score unaffected by gender or
protected classes. race.
Project POC: Camille Francois Project POC: Divya Tyam
The
Production Production
Sma Reply
Infrastructure
Mitigate unwanted and unwarranted subgroup
Bias analytics tools for use in vision inferences in suggested replies.
model development process Project POC: Sushant Prakash
If you'd like to discuss a project that you or your team are embarking on, have
a proposal you'd like to discuss, or have an interest in getting involved as a
contributor, please contact us at ml-fairness-leads@.
The
ML
Fairness
Project
GET INVOLVED
https://ml-fairness.googleplex.com/vanguard.html 3/3