10/05/2017
11:30 am - 12:00 pm
OCCC W230C

Objective: Guidance
Audience Level: Beginner/Intermediate
Session Type: Presentation

Machine Learning systems by default optimize for their dominant population at the expense of their peripheral users - what can we do to mitigate this? With impacts ranging from the performance of consumer applications to criminal sentencing, the effects of bias in ML will only grow in the next 10 years. This talk explores how Pinterest drives growth by explicitly designing its algorithms to acknowledge the heterogeneity of its users.

Speaker(s)