Strategic Classification
Add to your list(s)
Download to your calendar using vCal
If you have a question about this talk, please contact jg801.
When machine learning algorithms are deployed, we often assume their
predictions do not affect the outcomes they are trained to forecast.
However, this principle can break down when welfare (employment,
education, health) of rational individuals is affected. Knowing
information about the algorithm, such individuals may manipulate their
attributes in order to obtain a personally beneficial prediction (e.g.,
a higher credit score rating).
This reading group focuses on whether and how we can design machine
learning algorithms which achieve non-trivial performance even in
presence of strategic individuals. We further explore the potentially
negative social effects of strategic classification, and the broader
phenomenon of performativity of which strategic classification is a
special case.
This talk is part of the Machine Learning Reading Group @ CUED series.
This talk is included in these lists:
Note that ex-directory lists are not shown.
|